Social Commerce and Marketing

SOCIAL COMMERCE AND MARKETING

The point of least prestige

Written 16 September 2026, Hartwell, Northampton

THE POINT OF LEAST PRESTIGE

It is a Thursday in Hartwell and the light is wrong.

That is the first thing you learn about glass. You cannot see whether a window is clean by looking at it. You have to look along it, at an angle, with the light behind you, and even then the pane will tell you one thing from the path and another from inside the room. A window that is spotless at nine in the morning has a film on it at four in the afternoon because the sun has moved. The work is not difficult. Knowing whether you have done it is difficult.

There is a ladder against a gable. There is a pole system in the back of the van, and a tank of filtered water, and a set of brushes that get replaced more often than anybody who has not done this would guess. There is a sheet with a round on it, streets, houses, a six week cycle, and the sheet is the most valuable thing this business owns.

This is a document about a marketplace, and it is written from a ladder.

We want to deal with the obvious objection immediately, because it is a reasonable one, and because the answer to it is the reason all six of these articles exist.

The objection is this. Why would anybody take a window cleaner's account of how markets, platforms, ratings and marketing actually work?

The answer was written in 1958 by a sociologist at the University of Chicago named Everett Hughes, and it is not an answer about kindness or dignity or giving the working man his due. It is an argument about method. About where you should stand if you want to see how something works.

Hughes had spent years studying occupations at the bottom of the prestige ladder. Janitors, junk peddlers, people whose jobs are not on anybody's list of careers. And he had come to think that the value of that work was not what he first assumed. He wrote this, on page forty nine of Men and Their Work.

I have now come to the belief that although the problems of people in these lines of work are as interesting and important as any other, their deeper value lies in the insights they yield about work behaviour in any and all occupations. It is not that it puts one into the position to debunk the others, but simply that processes which are hidden in other occupations come more readily to view in these lowly ones. We may be here dealing with a fundamental matter of method in social science, that of finding the best possible laboratory for study of a given series of mechanisms.

And then the sentence that gives this article its title, from pages forty eight and forty nine.

It is a fruitful thing to start study of any social phenomenon at the point of least prestige. For, since prestige is so much a matter of symbols, and even of pretensions, however well merited, there goes with prestige a tendency to preserve a front which hides the inside of things. A front of names, of indirection, of secrecy. On the other hand, in things of less prestige, the core may be more easy of access.

Read that as what it is, which is a claim about visibility.

Prestige builds a front. A front hides the machinery. The higher the standing of a trade, the more of its actual working is concealed behind names, procedures, discretion, and the general reluctance of successful people to describe how they really operate. That is not dishonesty. It is what prestige is made of.

Down here there is no front. There is nowhere to put one. Everything this business does happens in the street, on a ladder, at a door, in view of the neighbours. When something goes wrong it goes wrong in public. When a system behaves badly towards us, it behaves badly in a way that can be described in full, because there is no reputation requiring it to be softened.

On Hughes's argument, a window cleaner writing about how the small business marketplace actually works is not an odd person to be doing it. He is standing in the better laboratory.

Hughes adds one more condition, on page forty eight, and we are going to try to hold to it throughout.

Until we can find a point of view and concepts which will enable us to make comparisons between the junk peddler and the professor without intent to debunk the one and patronise the other, we cannot do our best work in this field.

No debunking. No patronising. Nobody in this article is a villain and nobody is a hero, including us.

So, what this document is, and what it is not.

It is an opinion. We think a particular thing about the marketplace being described to small trades, and Section Fifteen says what it is plainly.

It stands on evidence, and the evidence is other people's. Sociologists, economists, statisticians and market researchers, named, dated, page referenced, and, where we have quoted them, quoted in their own words rather than ours.

And the join between those two is left visible. Every time we move from something that has been measured to something we believe, we say so in the sentence where it happens. That is the single rule this article is built on. Where we have broken it, Section Thirteen records it.

It is not advice. We are not telling any other business what to do.

It is not a complaint. Several sections describe arrangements that cost this trade money and standing. None of them alleges wrongdoing by anybody, and where we have no documented conduct to attach to a name, we name nobody.

And nothing in it comes from a job. No customer appears in this document. Not named, not anonymised, not with the details changed, not as a composite. Everything we know from fourteen years of standing outside other people's houses stays outside this document. Where we needed an illustration, we have used a published one from somebody else's research, and Section Eight explains exactly why.

Five articles before this one audited a subject each. Cybersecurity and sustainability, financial resilience, reviewing culture, chargebacks, and artificial intelligence. This one audits the marketplace itself. The claim, put to every one van business in Britain, that there is now a market of a particular shape, that you must buy your way into it, and that the alternative is to disappear.

Underneath all of it runs one finding, borrowed from a historian of science called Theodore Porter, and it is this. A trade that ran on personal knowledge and local reputation is being made to run on numbers instead. And when numbers replace judgement, that is not progress. It is a symptom. Of distrust, and of the weakening of the community that used to do the judging.

Everything that follows is that argument, tested against a round.

ONE. A NUMBER WITH NO PARENT

We went looking for the size of the UK social commerce market, the thing every small business is being told it must get into. Here is what came back, all of it for roughly the same year.

From eMarketer, eleven point seven five billion pounds for 2026.

From retail research, over twenty four billion pounds, heading toward forty billion pounds by the end of the decade.

From a market research vendor, eighty point five three billion US dollars for 2025, rising to one hundred and one point six nine billion US dollars for 2026.

From a second market research vendor, three hundred and seventy three point zero two billion US dollars for 2026, reaching five hundred and twenty three point six two billion US dollars by 2031.

The lowest and the highest of those differ by a factor of roughly thirty.

They cannot all be measuring the same thing. Some are almost certainly counting influenced transactions rather than completed ones. Some are probably counting the whole of social advertising. At least one looks like a global figure wearing a UK label. We do not know which is which, and we are not going to pretend we do.

We are publishing all four anyway. Not because we know which is right, we do not, but because a small business making a real decision about real money is being handed a thirty fold spread and told that it is data. If we picked one and printed it, we would be making exactly the choice that produced this mess. Selecting the number that suits the argument and presenting it as though it were the only one.

And there is one more thing to say about those figures, which we will say once and then apply to everybody including ourselves for the rest of the article. Every one of them is published by an organisation that sells reports or services into the market it is measuring.

That does not make them wrong. It makes them interested. Section Three has a name for what happens when a measurement is produced by the people who benefit from what it shows, and the name is not fraud. It is much more ordinary than that, and much harder to correct.

If the market size figures are the disease at scale, here it is in miniature.

TikTok says more than three hundred thousand UK small businesses sell through TikTok Shop. In the same set of results, for the same metric, the figure two hundred thousand also appears.

That is one platform, describing its own business, in two incompatible ways, in material that surfaces together.

We have not been able to establish which is current. Both are here. One hundred thousand businesses is the difference between them, which is to say that the discrepancy alone is larger than most of the markets anybody in this trade operates in.

We are not alleging dishonesty. Figures get updated. Old pages linger. A definition changes and the previous number carries on circulating. That is the point. Nobody has to lie for a number to become unreliable. It only has to be repeated by people who have no way of checking it and no reason to.

Fairness requires this next part, and it goes here rather than in Section Thirteen, because a rule that only applies to other people is not a rule.

An earlier version of this research recorded that internet sales were twenty seven point four per cent of all retail sales in Great Britain in July 2026.

The Office for National Statistics figure is twenty eight point three per cent. Ours was wrong.

It was checked, it was found to be wrong, and it has been corrected. It is logged here rather than quietly overwritten, because a document whose entire argument is about the traceability of figures does not get to silently fix its own.

There is exactly one figure in this field that has a neutral producer, a consistent methodology, a monthly publication schedule and an open licence. Internet sales as a share of all retail sales, Great Britain, July 2026. Twenty eight point three per cent. That is the Office for National Statistics, series J four M C, internet sales as a percentage of total retail sales. The bulletin is Retail sales, Great Britain, July 2026, released on the twenty first of August 2026, under the Open Government Licence version three point zero.

That is what a usable figure looks like. It has a series identifier. You can go and get it. If we have misread it, you can prove it in under a minute, which is the property that none of the four market size figures possess.

Two things follow from it.

First, roughly seventy per cent of retail spending in Great Britain still does not happen online at all. And this is retail, the most digitised category of consumer spending there is. For a mobile trade selling a physical service, performed at a physical address, by a person who has to travel there, the online share of the relevant market is lower still. Nobody has ever cleaned a gutter through a phone.

Second, and less comfortable for the story everyone is told. The same measure was twenty nine point two per cent in June 2026. Over that month the online share fell. Online sales values were down three point nine per cent month on month, having been up two point five per cent in June.

The longer run is upward. Online values were six point five per cent higher than July 2025, and eleven per cent higher across the three months to July than the same three months a year earlier. We are not claiming online retail is in decline. It is not.

We are pointing out something narrower. The one series in this field you can actually check does not move in a straight line, and the narrative sold to small businesses has no room in it for a month that goes down.

Here is the question that bothered us. These figures are not hidden. Anybody could lay them side by side in an afternoon, as we did. Why does nobody?

Part of the answer is in a 1999 paper by Sarah L Jack and Alistair R Anderson of the University of Aberdeen, writing about entrepreneurship education. They are describing why the enterprise agenda is so rarely subjected to scrutiny, and the first reason is that nobody can pin down what is being talked about. This is from page one hundred and fifteen.

Entrepreneurship has become an omnibus word, an overloaded concept and one that carries a variety of meanings which pundits pick and choose from at will. Its meaning ranges from small business operation, through new venture creation to innovation and individualism. As a heuristic the breadth of the concept allows it to be used in diverse rhetoric, explanation and justification.

A word broad enough to mean anything cannot be measured, because there is nothing stable to count. And a term that allows it to be used in diverse rhetoric, explanation and justification is not a failed definition. It is a working tool, and it is working exactly as intended.

The second reason is about atmosphere. From page one hundred and fourteen, they describe a rationale which is often uncritical, and which appears to operate at almost an ideological level of taken for granted wisdom.

You do not audit the taken for granted. There is no moment at which it occurs to you that the thing might need checking, because checking is what you do to claims, and this has stopped presenting as a claim.

And then the sentence that answers the question properly. Jack and Anderson report it from Goss, 1991, on page one hundred and sixteen.

Goss, however, rather cynically notes that the enterprise culture's affection for small business is rooted as much in the latter's intangible qualities as in precise measures of its effectiveness.

Read that slowly, because it is the load bearing sentence of this section.

The enthusiasm for small business is founded on what small business is felt to represent, as much as on any measurement of what it achieves. Grit. Independence. Self reliance. The person who had a go. Those are not findings. They are qualities, and qualities are not audited.

Nobody checks the figures because the figures are not what the enthusiasm rests on. A compliment does not have a methodology, and this trade has been receiving a compliment for forty years.

Jack and Anderson detect one further thread in all this, which they name themselves, and we are giving it one sentence and moving on, because the rest is an argument about post war capitalism and would turn this into a different article. They see in the enterprise agenda the production of what they call the friendly face of capitalism. An ideology that had taken damage in the post war years, finding in the small firm something likeable to be represented by.

That phrase is Jack and Anderson's, not Goss's. We have made that mistake ourselves in our own working files and corrected it. See Section Thirteen.

There is a more general version of what is going on, and it comes from Theodore Porter, a historian of science whose 1995 book Trust in Numbers asks why quantification spreads. Why so many decisions that were once made by judgement are now made by measurement.

His answer is not the expected one. It is not that numbers turned out to be more accurate. On page four he writes that where a consensus of experts is hard to reach, or where it does not satisfy outsiders, mechanical objectivity comes into its own. And on page eight.

A decision made by the numbers, or by explicit rules of some other sort, has at least the appearance of being fair and impersonal. Scientific objectivity thus provides an answer to a moral demand for impartiality and fairness. Quantification is a way of making decisions without seeming to decide. Objectivity lends authority to officials who have very little of their own.

Quantification is a way of making decisions without seeming to decide.

That sentence runs underneath every remaining section of this article, so it is worth being precise about what it claims. Porter is not saying numbers are bad. He is saying that the spread of quantification into a field is evidence about the condition of trust in that field. That numbers get reached for where judgement has stopped being accepted, and that a decision rendered as a measurement acquires the appearance of not having been anybody's decision at all.

Note his exact word, because it matters, and we had it wrong ourselves for a while. Not impartial. Impersonal. Impartial means unbiased. Impersonal means there is no person there. That is the property being purchased. Not accuracy, but the disappearance of the decider.

Hold that until Section Nine, where a five star average turns out to be doing precisely this job for a trade that used to be judged by neighbours.

We want to end this section with a man who spent a career counting things, because his caution is worth more than anybody's enthusiasm.

Robert Putnam's Bowling Alone is one of the most cited works of social science of the last thirty years, and it is built on an enormous quantity of membership data. And Putnam is careful, repeatedly, about what that data can bear. That nominal membership is not the same as involvement, and that a count of who belongs to something may be measuring the wrong quantity entirely.

The most quoted quantifier of community decline, warning that his own instrument may not be measuring the thing he cares about.

And here we have to declare a limit of our own. Our copy of Bowling Alone has a broken font map. The text is recoverable. Every numeral in it is not. Decoded sentences read, between the early s and the early s.

So we have used Putnam's argument and not one of his figures, and we could not have used his figures if we had wanted to. That is stated here, in the article, rather than buried. And there is an accident in it that landed the right way up, because the source we cannot take numbers from is the source warning us about numbers.

None of the above tells us how many houses to knock on. So here is the arithmetic this business actually runs on, offered as the check.

A residential street in a village like this one has somewhere between twenty and sixty houses on it. A working day gets through a certain number of them, and the number depends on access, height, frames, whether there is a conservatory, and whether the gate is locked. A gutter clearance is priced by the metre and by the height, and by whether there is a lightwell or a flat roof in the way. A render clean is priced by the elevation and by what has grown on it. Block paving is priced by the square metre and by how much has grown into the joints.

Every one of those numbers is checkable by the customer standing next to us. That is not a boast. It is the ordinary condition of a trade that works in front of people. A price for a job on a specific house is the least abstract number in this entire article, and it is the only one in it that anybody is going to be asked to pay.

Set that against a thirty fold spread on the size of a market.

The question this article is asking is not whether the online market is large. It is whether anybody reporting its size has any idea, and whether a business that cannot check a figure should be reorganising itself around it.

TWO. FOURTEEN YEARS OF KNOCKING ON DOORS

This is the control group.

Everything after this section describes a business operating inside a system of ratings, platforms, search results and paid introductions. This section describes the same business operating with none of them, for fourteen years, and it is here because you cannot say what a system changed unless you can say what was there before it.

The first round, in 2001, was one hundred and twenty houses.

Six pounds a house. Three pounds for conservatory windows, and that did not include the roof. It still does not, and the distinction has caused more confusion over twenty five years than any other single thing in this trade.

The tools were traditional. Squeegee, T bar washer, bucket, ladders, extension pole. Washing up liquid for detergent. The water was soapy and it made bubbles, and anybody who watched you work could see exactly what was happening and why.

No water fed pole. No purified water. No tank in a van, because there was no van.

There was a shopping trolley.

It lasted somewhere between six and twelve months, which was the length of time it took to pass a driving test and buy a car. There was no arrangement with anybody about the trolley. It was taken, it was used, and when it was finished with it was returned to the retailer it belonged to.

We are including that detail because leaving it out would make the beginning of this business sound more organised than it was, and because the gap between a shopping trolley and a filtration system is the entire distance this article is trying to describe.

Why this work. Not a calling, and there is no point pretending otherwise.

Easy access. Nobody's permission was required to start. No boss. Almost no overhead, a bucket, a squeegee, a ladder, a trolley. And you were earning the same day, which matters more than anything else on that list when there is no money.

And one thing that was not about money. The alternative. Other jobs had been tried. What they had in common was managers and supervisors who were mostly rude, positional power struggles that had nothing to do with the work, and tasks that were both demeaning and not difficult, which is an unpleasant combination, because there is nothing in it to be good at.

Window cleaning meant working with real people rather than inside a working environment. That was the trade off, and Section Eight is about what it cost.

Here is the script that was used at doors, unchanged for years, in the words it was said in.

Hello, I am Matt and I run Shining Windows. Do you have a window cleaning service provider? If not, would you like your windows to be cleaned, and how much would you like to pay?

Read the last clause again. And how much would you like to pay.

There is no marketing textbook in the world that recommends that. It hands the price to the customer before any relationship exists, before anything has been demonstrated, at the precise moment when conventional advice says to anchor high. It is, by every standard measure, the wrong thing to say.

It worked, and it worked for years.

We think it worked for a reason the rest of this article is about. In 2001, in these villages, the constraint was not price. It was availability. Asking what somebody wanted to pay was not weakness. It was an accurate reading of a market where the customer's problem was not cost but the impossibility of finding anybody at all.

It also does something else, which took us a long time to see. It makes the customer commit to a number out loud, in their own voice, at their own door. Whatever they say next, they have participated in setting it. Section Nine describes a system in which four firms are handed an address simultaneously and race each other downwards. The sentence above is the exact opposite arrangement, and it was arrived at by a twenty something with a shopping trolley who had never read anything about pricing.

What did a refusal sound like? No thank you. That is it. That is the whole of it.

Relaxed, ordinary, and not a rejection of a person. Most people said yes, because there were very few service providers around. That is a recollection of one area over one period and not a count, and we are marking it as such.

But the texture of the no is the thing worth recording, because it is gone. It was possible to knock on a stranger's door in 2001 and be turned down pleasantly. Section Four is about what happened to that.

A round that paid a week's wages took a few days, or a week of leafleting.

That sentence will look implausible to anybody who has started a service business in the last ten years, and it should. It is the single clearest measure of the difference between the two economies this article is comparing.

After that it stopped being door knocking. Phone calls and texts flooded in. Requests arrived without being asked for, which is Section Three's ninety per cent finding happening in real time in a Northamptonshire village, twenty four years before we read the research that named it.

So what were those customers actually judging? This is the question the section turns on, and the answer is not what we expected.

They were judging their windows.

Not credentials, not reviews, not a portfolio, not a brand. They looked at their own glass, formed a view that something ought to be done about it, and had discovered that doing something about it was surprisingly hard. The binding constraint was finding anyone at all.

So the first thing being judged was availability, and availability was scarce.

What they judged afterwards is the part that matters, and it comes down to four things. Fair pricing, flexibility, regularity, and the quality of the work. Not one of those is visible before the first job. Every one of them can only be assessed by having the work done and watching what happens next.

That is the whole of the analogue mechanism, stated plainly. You could not be judged in advance, because there was nothing to judge in advance. You were judged on the second visit, and the eighth, and the fortieth.

And one specific thing did the most work of all. The frames.

Most window cleaners cleaned the glass and left the frames. That was standard, it was widely accepted, and twenty five years later it is still what most of them do. Cleaning the frames took longer, cost more in time, and could not be charged for separately.

Once people had seen it, they wanted it. The round grew on it.

Consider what kind of information that is. It is not a claim. It is not a promise, a certification, a star rating or a guarantee. It is a physical difference, visible on a house, from a pavement, by anybody walking past, and it could be compared against the house next door without either party saying a word.

That is what a reputation was made of before there was a mechanism for reputation. Work that could be seen, on a street, by people who lived there.

And what it cost. February. No furlough. No sick pay. No reserve. When the weather made the work impossible, the answer was to sit indoors and wait for it to clear.

And here is the part we would not have believed if we had not been told it directly. There were no complaints. People rang up to say that the weather was bad, and that he should stay indoors, keep himself warm, and stay safe.

In his words. People used to care for me as an individual.

We are quoting that because it is the most important sentence in this section, and because it will be tested, twice. In Section Four the same trade is being told its rain policy is unacceptable. In Section Ten a stranger who has never bought anything is owed a reply within the hour.

Whatever changed between 2001 and now, it was not the weather.

The business has run since 2001. Across all of it there has been exactly one period of nearly stopping, and it was not commercial. It was music. Playing it, enjoying it, being in his twenties, having friends and a social life and living well.

Nothing else has ever put the tools down. Not a bad year, not a quiet January, not a lost round.

Snow, storms and rain stop the work. They always have.

And health stops the work. That has become a larger factor with age, and it is a fact about this business that a reader is entitled to know. The person who does the work is disabled.

That sentence is here once, stated as fact, and it is not an appeal and not an excuse. It is in this document for the same reason the price of a conservatory window is in it. Because an account of how a business actually operates that leaves out its actual constraints is not an account. It is advertising.

Section Four records what happens when that fact meets a system that has no field for it.

Four things are established here, and the rest of the article is measured against them.

One. A local trade could be built, from nothing, with hand tools and a borrowed trolley, to a week's wages in a week, because the constraint was availability rather than competition or cost.

Two. Nothing about the provider could be assessed in advance. Judgement happened after the work, repeatedly, over years, and it was made by people who could see the result on their own house.

Three. The differentiator was physical and visible from the street. Not a claim about quality. Quality itself, in a form a neighbour could check.

Four. The relationship ran in both directions. The customer was owed clean windows. The window cleaner was owed a phone call telling him to stay indoors when it was dangerous.

Every one of those four has since been replaced by something, and the next eleven sections are about what by.

THREE. HOW THE WORK ACTUALLY ARRIVES

Section One was about numbers with no parent. This section is about the opposite problem. A finding with a very clear parent, repeatedly measured, entirely uncontroversial, and almost completely absent from the material that tells small businesses how to market themselves.

The finding is that small firms get their customers by recommendation, and they always have.

That is not a slogan and it is not our opinion. It has been measured, by academics, in Britain, on samples large enough to take seriously. The strange part is not the finding. The strange part is what happened to it afterwards.

Our source is a paper by David Stokes and Nicholas C Wilson, both then of Kingston University, published in the International Journal of Entrepreneurship and Innovation Management under the title, Entrepreneurship and marketing education, time for the road less travelled.

They are not primarily reporting new data. They are reviewing what is already known and asking why the teaching of marketing does not reflect it, which is why this paper is useful to us, and why we are going to be careful throughout to name the researchers who produced each figure rather than crediting them to the two men reporting it. This is from their manuscript pages eight and nine.

One study of three hundred plus owner managers, Stokes and others, 1997, found that over ninety per cent of owner managers of small businesses used word of mouth communications as their primary way of transmitting marketing messages. Recommendations from existing customers were the most frequently claimed type of promotional activity. Moreover, the second most popular method of customer acquisition was recommendations from other sources, that is, non customers.

Three things in that passage, and they are separate claims.

Over ninety per cent, as the primary method. Not used it, not valued it. Primary. More than three hundred owner managers, and for nine in ten of them, the main way marketing messages travelled was one person telling another.

Recommendations from existing customers were the most frequently claimed promotional activity of any kind. Ahead of advertising. Ahead of everything a marketing budget would normally buy.

And second place went to recommendations from people who were not even customers. That is the one worth sitting with. The two top methods are both other people talking, and the runner up is other people talking who have never bought anything. The neighbour who has never used us but has watched us work on the street is, on this evidence, a more significant channel than most paid advertising.

Note the parent. That ninety per cent figure is Stokes and others, 1997, reported by Stokes and Wilson. It is a 1997 survey, which makes it old, and we are not going to pretend it is current. What it establishes is a baseline, and the reason we trust the baseline is what comes next.

The obvious objection is that this describes people with no ambition. One van operations who rely on word of mouth because they cannot afford anything else. That objection was tested. From manuscript page nine.

In a more recent study, of established small firms of more than five employees, nearly two thirds of the companies identified as high growth claimed recommendations to be the most effective way of attracting customers. Blackburn and Stokes, 2003.

Firms with staff. Firms identified as growing. Nearly two thirds of them saying the same thing.

Parent again, and it is a different one. This figure is Blackburn and Stokes, 2003, not Stokes and Wilson. Two separate research teams, six years apart, reported in one paragraph of one paper. We are separating them because Section One would insist on it.

So. A method used by nine in ten small firms, rated most effective by two thirds of the fast growing ones, measured repeatedly across sectors and sizes. Now look at what the marketing profession does with it. Again from manuscript page nine.

The discussions on word of mouth total about one page of the seven hundred and eighteen pages in Kotler, 2003. Brassington and Pettitt, 2006, run to a few paragraphs in over one thousand pages, with no listing of word of mouth in the glossary or index.

Kotler is the standard marketing textbook, the one that sits on the reading list of most marketing degrees in the English speaking world. One page in seven hundred and eighteen.

Brassington and Pettitt is the standard British one. A few paragraphs in over a thousand pages, and it is not in the index. A student could read that book cover to cover and never encounter, as a named topic, the thing nine out of ten small firms actually do.

It gets better. Of the twenty eight courses advertised by the Chartered Institute of Marketing on marketing communications, none of them major on word of mouth.

And, finally, the one that ought to be embarrassing. Business Link have a benchmark index that claims to be an extensive benchmarking resource for small business. However, the questionnaire and notes that create the input for the process contain no mention of word of mouth, or stimulating recommendations or referrals.

A government backed benchmarking tool, built specifically for small businesses, which does not ask about the single thing small businesses overwhelmingly report doing.

Stokes and Wilson's own conclusion is measured. There is, they write, a mismatch between what entrepreneurs actually do in order to acquire customers and what the discipline teaches.

We would put it less politely, and we will mark that this is our view and not theirs. The apparatus that sells marketing to small firms has a commercial reason to under describe the method that costs nothing. We cannot prove that and we are not asserting it as fact. We are saying the incentive exists, it points in exactly that direction, and Section One's rule, ask who benefits from the number being reported, applies to absences as well as to figures.

This next part changed how we think about our own business, and it is a straightforward empirical correction to something almost everybody in the trade believes.

Ask a business owner which customers refer the most, and they will tell you it is their best ones. The long standing, loyal, been with us for years customers. Stokes and colleagues found exactly that belief in 2002. From manuscript page ten.

Although this appears logical and the accepted wisdom of many business owners, it is often the opposite that is more accurate. A number of studies, for example East and others, 2000, have shown that it is the most recent customers that refer most frequently. They are enthused by positive recent experiences that they are keen to share with friends.

And then the sentence that explains why. Thus, it is stories that make recommendations, and more recent experiences tend to be the stories that we pass on the most.

That is cited at second hand. East and others, 2000, is the study. Stokes and Wilson are reporting it. We have not read the original and we are saying so.

It is stories that make recommendations. That is the load bearing sentence of this section, and possibly of the article.

A long standing customer of nine years has no story. Nothing happened. The windows are clean, the way they have been clean since 2017, on a Thursday, as expected. There is nothing to tell, and that absence of narrative is not a failure of the relationship. It is the relationship working perfectly.

A customer whose gutters were cleared last week during the first proper storm of the autumn has a story. It has a beginning, a problem, a person arriving, and a resolution. It survives being retold over a fence.

Loyalty and referral are not the same variable, and this trade has been conflating them for decades. Us included.

Now take that finding and put it next to the thing every small firm is currently told to do, which is to accumulate ratings.

A rating is not a story. It is what is left of a story after the story has been compressed for transmission across a system that cannot carry stories.

Five stars. Four stars. Great job, very tidy. The narrative, the storm, the ladder, the thing that was blocked, the person who turned up, is precisely the part that does not fit in the field, and it is precisely the part that East and colleagues found does the work.

The channel that small firms are being urged into is optimised to strip out the ingredient that makes referral function.

We are not going to develop that here, because it is Section Nine's whole subject. But this is where it begins, and it begins with a 2000 study about which customers refer, not with anybody's opinion about platforms.

There is a deeper reason underneath it, and it comes from Theodore Porter, whose book Trust in Numbers has been running underneath this article since Section One. Porter's subject is why quantification spreads, and his answer is not the one most people expect. It is not that numbers are more accurate. On page four he writes that where a consensus of experts is hard to reach, or where it does not satisfy outsiders, mechanical objectivity comes into its own.

Read that against a recommendation.

A recommendation from a neighbour is the opposite of mechanical objectivity. It is personal, unverifiable, non standardised, and entirely dependent on your relationship with the person making it. It cannot be audited and it does not scale. By every criterion Porter lists, it is a worse instrument.

And it is also, on the evidence above, what nine in ten small firms actually run on.

Porter's finding is that the substitution of numbers for judgement is not a sign of progress. It is a symptom. Of distrust, and of the weakening of the community that used to do the judging. A rating system is not an upgrade on a neighbour's word. It is what you build when there is no neighbour.

That is the argument of this entire article, and it is why Porter is in it.

We should show our own working, because a section arguing for transparency that does not disclose is worthless.

The window round grew street by street. Not by area, not by postcode targeting, not by leaflet drop. By one house on a street, then the house next door, then the one opposite. That is the mechanism the 1997 study describes, observed from inside, and it is slow and it compounds.

The render work comes from being watched. Render cleaning is the most visible thing we do. A rendered gable that was green on Monday and is not on Wednesday is a demonstration, performed in public, at the client's expense. More of that work has come from neighbours who saw it happen than from anything we have ever said about ourselves.

The block paving is the same pattern, one street removed. A restored drive is visible to everybody who walks past it, and drives are what people compare.

Commercial work has come through residential customers. Somebody whose house we clean works somewhere that needs its windows done, and mentions it. That is East and colleagues' finding wearing a high visibility vest. The person who refers is the one who has recently had the experience.

And we have paid for leads. We are not going to pretend otherwise, and Section Nine is not written from the outside. Some of it converted and some of it did not.

Three things this section is not claiming.

It is not claiming that advertising does not work. It is claiming that the dominant method reported by small firms is under described by the discipline that teaches marketing, and that this is measurable and odd.

It is not claiming these figures are current. The primary survey is from 1997, the growth study from 2003, the textbook counts from 2003 and 2006. They establish a baseline, not a snapshot of 2026. Anybody who repeats them as current is doing the thing Section One condemns.

And it is not claiming our own experience proves anything. Fourteen years of one business in one county is an anecdote. It is placed beside the research because the two agree, and we would have said so if they disagreed.

FOUR. THE YEAR THE RULES CHANGED

We went into this section looking for a scene.

That was the plan. Find the single moment, a job lost, a quote undercut, one conversation, where fourteen years of goodwill were visibly outranked by somebody's advertising budget. One scene, told properly, would carry the whole transition.

There is not one.

We asked the question directly and the answer was, no, not a singular date or event. No job. No conversation. No moment where it turned.

That absence is the finding, and it is a better finding than the scene would have been.

Nothing was taken. Nothing was announced. There was no day on which the rules changed and somebody said so. The conditions of the trade were altered gradually, by accumulation, and the alteration was only visible in retrospect, which is precisely what you would expect of a change that nobody decided and nobody is accountable for.

Section Eleven is about that shape. This section is about the year it started.

2015. The realisation was that being findable now mattered more than being known.

Those two words are doing a lot of work and they are not synonyms. Being known is a state that exists in other people. Several hundred households who have an opinion about you because you have cleaned their windows for a decade. Being findable is a state that exists in a system, and it can be switched off by somebody you have never met.

The response was not passive. A website was built. A year was spent studying search engine optimisation and marketing, which is worth pausing on, because it is a self employed window cleaner taking a year to learn a technical discipline in order to keep doing the job he was already doing perfectly well.

And in 2015 it worked, and it worked in a way that has since disappeared. In his words. Back then it was easy to be human online. People saw your post with a picture of quality and hired me.

A photograph of good work. That was the whole mechanism.

It is exactly the mechanism from Section Two, work visible on a street, judged by people who can see it, relocated onto a screen and still functioning. For a few years the new channel carried the old signal.

What was built was an e commerce website with online payments and online appointments. That is not a brochure page. It is infrastructure, built in 2015, by a one van business in a village.

Then Facebook. And what went on it was not advertising. Free cleans, given to people who could not afford to pay. In his words, it spread like wildfire.

We want to be careful here, because this is the sort of thing that gets told as a heartwarming marketing lesson and it is not one. The free cleans were not a customer acquisition strategy. They were given to people who could not afford a window cleaner. That they also worked is true and is beside the point. If this is the moment in the article where somebody decides we are recommending free work as a tactic, we have written it badly.

The cost. About eighteen pounds a month for the website in 2015. It is twenty eight pounds and eighty pence a month now.

That is a real number with a real parent, us, and it is the only cost figure in this article we can state with complete confidence. It has risen about sixty per cent in a decade. We are not going to build an argument on it. It is simply what the thing costs, stated, because Section One's rule applies to us as well.

Three shifts happened in the customers, in order.

First, the opening question changed. In 2001 people rang to ask about availability. Could you come, and when. By the late twenty tens, people rang to talk about something somebody else had written about us.

That is a complete inversion of the mechanism described in Section Two. The judgement had moved from after the work, by the person who received it, to before the work, by a stranger.

Second, a new step appeared in the middle. People would ring, raise the review, work out in the course of the conversation that they were talking to a human being, and then book.

That step did not exist before, and it costs something every single time. Somebody arrives requiring to be persuaded of something that was never previously in question. That there is a person here. Section Ten has a name for what that step is made of, and who pays for it.

Third, the price expectation did not move. We expected it would. Comparison is supposed to drive prices down, and this is the finding that surprised us most. It did not. What people expected to pay stayed roughly where it was.

What changed instead was what they thought they were buying. Here is his account of what a customer in 2026 believes they are assessing.

Customers have no idea what they are paying for other than pure water, and everyone now uses a brush. That is the common perception and the qualifying data for consumers.

Pure water. A brush.

That is it. That is the entire technical vocabulary the market has acquired in a decade of being online. Two signifiers, both of which every operator in the country now possesses, neither of which distinguishes anybody from anybody.

Compare it against Section Two. In 2001 the thing being judged was the frames. A physical difference, on a house, visible from a pavement, checkable against next door by anybody walking past. In 2026 the thing being judged is whether the van says pure water on it.

The trade did not get less skilled. The available vocabulary for describing skill collapsed, and it collapsed into exactly two terms, both of which are inputs rather than outcomes. Nobody is judging the result any more. They are judging the equipment.

That is Porter's substitution arriving in a form nobody would recognise as quantification. A number is not required. A pair of standardised labels will do the same job, provided everybody agrees to use them and stops looking at the windows.

A specific kind of phone call became common, and it opens the same way every time. Last cleaner charged five pounds. Can you beat it.

There is one question that sorts this call, and it takes four seconds. When were your windows last cleaned?

I do not know. Or, about a year ago.

That answer tells you that the previous arrangement was not a service that lapsed. It was a price that was shopped, once, some time ago, and is now being shopped again.

And, this is his account, from twenty five years of it, that customer complains. Reliably. And that customer is frequently the one behind closed curtains in the evening, not answering the door to the sole trader who has come for his three pounds. Or his six. Or his twelve, depending on the era.

We are stating that as this business's experience and not as a finding about people in general, because it is an impression formed over years and not a count, and because Section One does not let us dress a pattern up as data.

But there is something structural underneath it, and it is not about anybody's character. Shopping a price and buying a service are two different transactions, and the second one, the one Section Six describes, the recurring arrangement with a specific person, is the one the current channel is worst at producing.

The cost of doing this work has risen sharply. Fuel, primarily, and everything else that has risen with the cost of living. The going rate has not moved with it. Plenty of operators are still charging twelve to fifteen pounds.

Two things follow, and he has watched both happen.

Work gets concentrated geographically. If the rate will not move, the only remaining variable is travel, so rounds contract into tighter and tighter local areas to strip out the cost of getting there.

And the service range contracts too. Most providers start with a full set of tools. The gutter kit, the pressure washer, the render equipment. Things break. Replacing them costs money the rate no longer generates. So they stop offering those services and end up doing windows only, not by choice but by attrition.

The evidence for that second pattern is trade forums, trade posts, and conversations with other operators over two decades. That is not a survey and we are not presenting it as one. It is the accumulated impression of somebody inside the trade, offered as exactly that.

But note what it describes, because it is Section Nine's argument arriving from an unexpected direction. A market that cannot support tool replacement will produce operators who can only sell the one service that needs the least equipment. Nobody decided the trade should narrow. It narrowed because the rate stayed still while the costs did not.

So there was no single moment. Here is what there was instead, and it is his own account.

Five years in which the public record of this business was substantially written by people who had never bought anything from it. Reviews from people who had not had a service. Statements about who we are and what we do that are not true.

And a set of things the format has no room for. That we do not work in the rain. That a hospital appointment cannot be moved. That the person doing the work is disabled. None of those is a defence available in a review box, because a review box does not take defences. It takes a number and a paragraph, and the paragraph is not addressed to you.

Underneath all of it, in his words, is something with no sociology attached to it at all. We want to feed our children and pay our bills.

This is Article Three's territory and we are not going to take it over. Reviewing Culture deals with the mechanics. Why a single negative outweighs many positives, why the people most motivated to write are the least representative, what can and cannot be done about a false review. That is where this belongs and that is where it is dealt with.

What belongs here is one narrower observation, which is this section's whole point. The transition was not that a competitor outspent us. It was that the thing being evaluated stopped being the work.

In 2001 the assessment was made by the person who received the service, on the evidence of their own house, repeatedly, over years. By 2020 a large part of it was made by people with no relationship to the service at all, in a format with no room for circumstances, and it arrived before the first conversation rather than after the fortieth.

That is not a marketing problem and no amount of advertising budget answers it. Section Nine explains why the mechanism exists. Section Eleven is about how often this shape recurs.

One last change, and it is the one that would have been hardest to predict in 2001. People stopped answering the door.

Not everybody, and not always. Plenty are at work, which is ordinary and always was. But in his account of the last three to five years, most people will not come to the door, including when the windows are open, including when they can see you and you can see them.

The trade's word for the resulting job is a blind clean. The work is done, nobody is seen, nobody speaks.

Set that against Section Two, where the entire business was built by knocking on doors and a refusal sounded like, no thank you. The same act now mostly produces no response at all.

We are not going to theorise about that, because we would be guessing. We are recording it, because it is a change in the physical conditions of this trade that no market report contains and no platform measures, and because the doorstep is where Section Eight happens.

FIVE. WHY A CLEAN HOUSE WHEN THE MONEY IS TIGHT

Here is something we could not explain for about ten years.

In a bad year, and there have been bad years, nationally and locally, the large jobs go first. The render clean gets deferred. The driveway restoration waits. The soffits and fascias can, apparently, wait indefinitely.

And the six weekly window clean, which is the smallest recurring line on any household's list of outgoings, mostly stays.

That is not what we expected when we started, and it is not what the trade folklore predicts. The folklore says the luxury goes first and the window cleaner is the luxury. It turns out to be more complicated than that, and there is a body of consumer psychology that explains part of it, with two large caveats that we are going to state before we state the findings, because we have already got this wrong once in public.

The first version of this article used something called the lipstick effect to explain the pattern above. The lipstick effect is the observation that spending on small cosmetic items appears to hold up, or even rise, during recessions. We used it to argue that a forty pound window clean is the lipstick of the house.

That was an over extension of the source and we have corrected it. The study we were leaning on, Hill, Rodeheffer, Griskevicius, Durante and White, in the Journal of Personality and Social Psychology in 2012, is a careful piece of work about something much narrower than we implied. Its own abstract says what it is.

Recessionary cues, whether naturally occurring or experimentally primed, decreased desire for most products, for example electronics and household items. However, these cues consistently increased women's desire for products that increase attractiveness to mates, the first experimental demonstration of the lipstick effect. This effect is driven by women's desire to attract mates with resources and depends on the perceived mate attraction function served by these products.

Five experiments. The mechanism is mate attraction. It is specifically about women, and the authors say so repeatedly and design for it. Stretching that to cover a household's decision about its guttering is not a small liberty. It is using an evolutionary psychology finding as a general purpose explanation for anything that survives a downturn, which is precisely the move that makes pop psychology useless.

So why is it still in this article? Because one part of it does apply, and it is the part we originally skipped over. Recessionary cues decreased desire for most products. The finding that spending under economic pressure is selective rather than uniform, that it does not simply scale down across the board but reorganises, is real, evidenced, and exactly as broad as the study supports.

That is what Hill and colleagues are doing in this article now. Not the lipstick. The selectivity.

The correction stays visible in this section rather than being quietly deleted, because we would rather be the sort of document that shows its working than the sort that is never wrong.

The model that does fit is set out in a review paper by Naomi Mandel, Derek Rucker, Jonathan Levav and Adam Galinsky, who pull together several decades of work into something they call the Compensatory Consumer Behaviour Model.

The starting point is a self discrepancy. A gap between how a person currently sees themselves and how they would like to. The term is Higgins's, from 1987. The idea is that consumer goods carry psychological value alongside their functional value, and that purchases get used to manage that gap.

The model identifies five things people do about a self discrepancy. All five are worth knowing, because three of them predict opposite behaviours, and that turns out to be the honest heart of this section.

One. Direct resolution. You address the actual source. Unhappy with your fitness, you join a gym.

Two. Symbolic self completion. You signal mastery on the same dimension as the gap, without addressing its source. The theory comes from Wicklund and Gollwitzer, and their own example is precise. Wicklund and Gollwitzer, in 1982, found that MBA students who lacked certain objective indicators of business success, for example a high grade point average or multiple job offers, were more likely to display other, symbolic indicators of business success.

Three. Dissociation. The opposite move. You avoid the domain entirely. Mandel and colleagues describe it as avoiding purchases in the domain of the self discrepancy.

Four. Escapism. You direct attention elsewhere.

Five. Fluid compensation. You affirm the self somewhere else entirely. In their words, in fluid compensation people address a self discrepancy by affirming the self in a domain distinct from the domain of the self discrepancy. The idea of fluid compensation is a core tenet of self affirmation theory.

We are stating that one carefully because we nearly got it wrong. Fluid compensation is about what a person does for themselves in an unrelated area. It is not, and this paper does not support, an explanation of how anybody treats anybody else. See Section Thirteen.

Then there is a sentence in the same paper that does more for this section than the rest of the model combined.

Symbolic self completion need not result in an increase in overall consumption as in the preceding examples. Instead, it can direct consumers to a specific set of options within a category without changing total consumption.

Read that against what we actually observe.

We do not see households spending more on their property in a bad year. We see them spending differently. The total falls, and within what remains the composition changes. The big visible transformation is postponed. The small recurring thing that keeps the front of the house looking maintained is retained.

That is not an increase in consumption. It is a redirection within a category, which is precisely what the model says symbolic self completion does.

And the study they cite immediately after is the one that names the mechanism. Cutright, in 2012, found that when individuals perceived a lack of personal control, they attempted to reassert control by choosing products that contained symbolic boundaries. A framed painting rather than an unframed one.

A frame. A defined edge, placed around something, by somebody who is not in control of the wider situation.

Hold that thought for exactly one section, because Section Seven is about what exterior cleaning physically is, and the answer turns out to be the restoration of edges. We would rather make that argument there, properly, than glance at it here.

Here is something the trade believes that our own experience does not support. The folklore says the customer who negotiates hardest on price is the customer with the least money. Fourteen years of doing this says that is not reliably true, and we have never had a way of accounting for it.

Mandel and colleagues, reviewing Rucker and Galinsky's experimental work, report this on manuscript pages seven and eight. Feeling less powerful than others can lead consumers to report a higher willingness to pay for high status products, in order to restore feelings of power.

The mechanism is not income. It is the experience of standing lower in a hierarchy, which is a different thing, and which comes apart from income constantly.

We are going to be careful about what that does and does not mean, but we will say what it makes available. It offers an account in which pressing hard on a price and then wanting a conspicuously finished result are not contradictory behaviours by a confused person. They can be the same motive arriving twice.

And we are not diagnosing anybody. This is a mechanism the literature says is available to everyone, ourselves emphatically included. It is not a description of a customer, a type of customer, or anybody we have ever worked for. No customer appears in this document.

Now the caveat we have been building towards, and it is the paragraph that makes this section worth writing rather than worth deleting. The model predicts both outcomes.

Symbolic self completion says a person under pressure may redirect spending towards the visible, the bounded, the maintained. Dissociation says a person under pressure may avoid the entire domain. Both are in the same paper, both are supported, and neither one tells you which house you are standing in front of.

Cutright's experiment is about paintings. Not houses, not render, not window cleaning. Wicklund and Gollwitzer's is about MBA students. Rucker and Galinsky's is about laboratory purchases. Not one of these studies has anything to do with the exterior of a home in Northamptonshire, and none of the authors would claim otherwise.

So here is what this section is doing, said flatly. We are placing a body of findings beside fourteen years of observation and saying that they appear to be about the same thing. We are not claiming the findings explain the observation. We are not claiming a causal link. We are not claiming anybody has tested this in our trade, because nobody has.

That sentence is the difference between this article and every marketing blog that has ever cited the lipstick effect at a small business. The blogs take a laboratory result and sell it as a technique. We are showing you the result, showing you what we see, and leaving the join visible so you can decide whether we have joined them too enthusiastically.

The observations, stated as observations.

The render clean gets deferred. It is the largest single job we do on a domestic property and it transforms a house. In a tight year it is the first thing to go, and it often goes for two or three years running.

The driveway waits. Block paving restoration is discretionary in a way that almost nothing else we do is. A green drive is not a problem. It is an appearance. Appearances wait.

The six weekly window clean stays. Smallest recurring cost, most frequent contact, and the thing most likely to survive a household's cull. Sometimes it moves to eight weekly. It rarely stops.

And gutter clearance does not wait, ever, and it is the exception that proves the rule. Because a blocked gutter is not an appearance problem. It is water going into a wall. Customers who have deferred everything else will book a gutter clear at short notice and pay for it without discussion, because the thing being managed is not how the house looks. It is fear, and fear behaves differently from every other motivation described in this section.

That last one is worth sitting with. Everything in the compensatory consumption literature is about identity, appearance, status and the gap between selves. None of it explains the gutter. The gutter is bought for a straightforwardly rational reason by people acting on a correct assessment of risk, and it is a useful reminder that most of what customers do is sensible and needs no psychology at all.

A model that explained every purchase would be a model we should not trust.

Three things this is not for, because this is the section most likely to be mistaken for a sales technique.

It is not a sales technique. Nothing here tells anybody what to say to a customer, and if it did we would have cut it. Knowing that a purchase may carry symbolic weight is not a licence to lean on the symbol.

It is not a claim about any individual. Every mechanism described is general, experimental, and available to everybody. It is not a way of reading a person on a doorstep, and anybody who thinks it is has misunderstood what a laboratory result can support.

And it is not an argument that the work does not matter functionally. Clean glass lets in light. Cleared gutters protect walls. Restored paving is safer underfoot. The psychology sits on top of work that is worth doing for plain physical reasons, and it does not replace them.

SIX. A ROUND IS A ROUTINE

Every section so far has been about how work is found, priced, counted or judged. This one is about something else, and it is the part of this trade that the entire marketing apparatus described in Section Three cannot see, because there is no field for it on any form.

A round is not a list of customers. It is a rhythm.

We want to try to say what that means, and we are going to need two sources who have nothing to do with cleaning to say it. One a sociologist writing about modern life in general, and one a pair of American researchers writing, in 1956, about people on television.

Anthony Giddens uses a term for something most people have and nobody notices. Ontological security. Matt Adams, in a doctoral thesis that is in large part a critique of Giddens, sets it out clearly, and we are following Adams here because he is the source we hold and because he does not take Giddens on trust.

The idea starts with what Giddens calls going on. The business of getting through an ordinary day. As Adams puts it, on page thirty seven of his thesis.

Going on refers here not just to the physical ability to continue of course, but to a conscious and emotional ability to carry out routines of speech, gesture, posture and thinking without being overwhelmed by uncertainty about their correctness or authenticity. More concisely, going on presumes a level of ontological security, which Giddens defines as a sense of continuity and order in events, including those not directly within the perceptual environment of the individual.

Read the second half of that definition slowly, because it is the part that matters here. Including those not directly within the perceptual environment of the individual. Continuity and order in events you are not currently watching.

You are not at home right now. You are fairly confident your house is still there. You are not thinking about it, and that is the whole point. Adams stresses that this work passes unnoticed, so deeply engrained is it in practical consciousness. It only becomes visible when it fails.

That background confidence is not free. It is made of things that keep happening in the same way, and it is assembled out of a very large number of very small reliabilities.

Giddens's argument, as Adams reports it on page sixteen, is that this used to be anchored in place, and increasingly is not. In traditional communities, time and space were essential aspects of the local ordering of social life. They were inseparable from considerations of place. Less than a hundred years ago, for the majority in the West, activities and relationships were largely determined by geography. And then the sentence. One's locality, Giddens writes in 1990, is the focus of, and contributes to, ontological security in ways that are substantially dissolved in circumstances of modernity.

Substantially dissolved. Not abolished. Dissolved, which is a much better word, because it describes something going into solution rather than disappearing. It is still in there. You cannot see it and you cannot get it out.

Adams then puts Giddens next to E P Thompson, who described how a pre industrial community organised its day. Not by the clock but by what needed doing. Thompson, writing in Customs in Common, cites the example of a crofting and fishing community.

Whose framework of marketing and administration is minimal, and in which the day's tasks, which might vary from fishing to farming, building, mending of nets, thatching, making a cradle or a coffin, seem to disclose themselves, by the logic of need, before the crofter's eyes.

Thompson calls this task oriented time. The work announces itself. Nobody schedules the thatching. The roof does.

Now hold that against a six weekly round.

The glass tells you when. So does the gutter, and the render, and the north facing block paving that goes green first every year without being asked. A round is a diary, obviously. We keep one, it is a proper one, and Article Five is about what keeping it costs. But underneath the diary there is a much older logic running, in which the work discloses itself by the logic of need, in front of your eyes, on a street.

Task oriented time, surviving inside a clock time economy. We are not claiming that makes this trade pre industrial or romantic or better than anything. We are pointing out that the structure Thompson described has not vanished. It has retreated into the handful of jobs still done outdoors, on foot, in a fixed geography, by the same person each time.

The other half comes from an unexpected direction. In 1956 two researchers, Donald Horton and R Richard Wohl, published a paper in Psychiatry about the relationship between television and radio audiences and the people they watched. They coined the term para social interaction for it, and the paper is now sixty nine years old and better than most things written since.

They were interested in why a broadcaster comes to feel like somebody you know. Part of their answer is about manner and address. But the part that stopped us is not about performance at all.

The persona offers, above all, a continuing relationship. His appearance is a regular and dependable event, to be counted on, planned for, and integrated into the routines of daily life. His devotees live with him and share the small episodes of his public life. Indeed, their continued association with him acquires a history, and the accumulation of shared past experiences gives additional meaning to the present performance.

Strip out the television. What is left is a description of recurrence, and of what recurrence does over time. A regular and dependable event. Counted on, planned for. Integrated into the routines of daily life. Acquires a history.

Put those two together, and the putting together is ours, not Giddens's and not Horton and Wohl's, neither of whom was talking about anything like this.

Giddens says ontological security is built from continuity and order in events, and that locality used to supply it and increasingly does not. Horton and Wohl describe what a reliably recurring appearance does to the person who experiences it.

They are describing the same structure from two directions. And a tradesman on a fixed round sits precisely at the intersection. A regular and dependable event, occurring in a specific locality, that has acquired a history.

We think this is the real answer to a question that has bothered us for years, which is why customers stay.

Not because we are cheap. We are not the cheapest and never have been. Not because the work is unrepeatable. It is not. Cleaning a window is not a rare skill and we would be lying to say otherwise. Any competent operator with a pole and a tank could do it.

What they could not do is have been coming for nine years.

What a long standing customer is buying is not only clean glass. It is the recurrence of an expected thing in a world where fewer things recur.

This is not theory from where we stand, so here is what it looks like.

There are customers who know which Thursday. Not the date. The position in the cycle. They could not tell you our schedule and they do not need to, because they have absorbed it the way you absorb the bin day.

There are side gates left unlocked on the morning we are due, in houses where nobody is home, by people who lock everything else. There is money left under a mat, or in a jar in a porch, or in an envelope with a name on it, in a village where the same people would not leave a bicycle out.

There are customers who ask us to come anyway when the glass barely needs it, and who are mildly embarrassed about it, and who should not be, because what is being maintained is the cycle, and the cycle is the product.

There are customers who have moved house within the area and taken us with them, which means the relationship was not with the building.

And there are conversations at doors. Not long ones. Two minutes, six weekly, for years. Over that period you learn who has been ill, whose son has gone to university, who has been widowed. None of that is in the diary and none of it is going in this document.

The round is the part of the business that survives everything else. Every quiet January, every broken bit of kit, every difficult week. It is the thing that is still there on the other side.

We want to record something here that has no commercial purpose whatsoever, because leaving it out would make this section a nicer piece of writing and a less honest one.

Customers die.

When you have cleaned the same house every six weeks for nine or twelve years, you find out in the way neighbours find out. Sometimes there is a note. Sometimes the curtains are wrong. Sometimes a relative answers the door and you understand before anybody says anything.

And what happens then is that a line comes off the sheet.

The commercial fact is minor. It is one house on a street of many, and there is a straightforward business response, which is that the round is rebuilt. That happens and it is fine. But there is a gap for a while, in an actual physical sequence of an actual physical street, and the gap is noticeable in the way a missing tooth is noticeable, and after enough years you have a small number of these.

Horton and Wohl's word for what accumulates is history. That is what it is. You cannot build fourteen years of recurrence without also building the losses that come with it, and any account of a round that describes only the good part is selling something.

We mention it for one reason and then we will stop. Everything Section Nine describes, the rating, the lead fee, the response time metric, the platform's model of what a tradesman is, has no field for this and could not have one. The thing that makes a round valuable is the thing that cannot be recorded in the system that claims to measure it.

This section is the one most likely to be over read, so we will mark the limits ourselves, clearly.

Giddens is a theorist, not a researcher of window cleaning. He is writing about modernity in the broadest possible terms. No study anywhere shows that a tradesman on a regular visit functions as an ontological anchor for anybody. We are not claiming one does. We are saying, here is an account of where a sense of continuity comes from, and here is a trade that supplies a recurring event in a fixed place over long periods, and the two appear to be about the same thing.

Adams's thesis is a critique of Giddens, not an endorsement, and we have used it precisely for that reason. He does not take the framework on trust and neither do we.

Horton and Wohl were writing about broadcasters. Their subject was a relationship in which one party appears and the other watches. That is not this. It is, as it happens, almost the opposite of this, and the difference between them is the subject of Section Ten, so we will only plant the flag here.

The difference is reciprocity. On a round it runs both ways. We turn up, they leave the gate open, the money is under the mat, there is a conversation at a door. Both parties are present and both parties do something. Horton and Wohl's viewer is present and the persona is not. What they are describing has the recurrence without the return.

A round has what a feed has lost. Hold that thought until Section Ten, where it is the whole argument.

One more thing this explains. Section Nine ends with a platform's review history being property held in somebody else's database. This section is the other half of that, arrived at from the customer's side.

A round is not held anywhere. There is no server. There is no account. It exists as a distributed arrangement in several hundred households, each of which independently expects a particular thing to happen at a particular interval, and each of which can end it without notice and for any reason.

That makes it fragile in a way a contract is not, and unconfiscatable in a way a platform account is not. Nobody can suspend it. Nobody can vary the terms. Nobody can decide that this year the algorithm favours a different sort of firm.

It can only be earned, and it can only be lost, and both of those take years.

SEVEN. THE FRONT OF THE HOUSE

This is the shortest section in the article and it contains the one observation in it that nobody but this trade could have made. Here it is up front, and then we will show our working.

What exterior cleaning physically consists of is the restoration of boundaries. Not surfaces. Edges.

Start with something obvious that is worth saying slowly. The outside of a house is not the house. It is the part of the house that is presented to everybody who passes, whether or not anybody inside is home, whether or not anybody wants to present anything. A living room is arranged for the people who use it. A frontage is arranged, or not arranged, which is also a statement, for a street.

It is the only part of a dwelling maintained for an audience the occupant did not choose and cannot exclude. That is the fact this section rests on, and everything else follows from it.

There is a research literature on the relationship between people and the places they live, and the standard collection is Place Attachment, edited by Irwin Altman and Setha M Low, published by Plenum Press in 1992 as volume twelve of a series on human behaviour and environment. Their opening chapter is careful about what the term means, and the care is useful to us. This is from page seven.

The term place attachment implies that the primary target of affective bonding of people is to environmental settings themselves. Thus many authors refer to places as satisfying because they permit control, creativity, and mastery, and they provide opportunities for privacy, personal displays, security, and serenity.

Read that list. Control. Mastery. Personal displays. Security. These are not decorative words. They are the reasons the editors give for why a place can be satisfying at all.

And then they immediately complicate it, which is why this is a good source rather than a convenient one.

On the other hand, a number of scholars indicate how attachment to places may be based on or incorporate other people, family, friends, community, and even a culture. Thus the social relations that a place signifies may be equally or more important to the attachment process than the place itself.

The attachment may not be to the place. It may be to what the place signifies about a person's relations with other people.

That sentence does a great deal of work here, because it explains something we see constantly and had filed under embarrassment.

Customers apologise for the state of the front of their house. Not sometimes. Routinely, and to a degree that is out of all proportion to anything that is actually wrong. Green render. A drive with moss in the joints. Guttering with grass growing out of it, which, it should be said, is extremely common, entirely normal, and the reason we have a business.

Nobody apologises to a plumber for a leak. A leak is bad luck. The outside of a house is felt to be different, and Altman and Low explain why. If the attachment to a place incorporates what the place signifies about one's standing with other people, then a frontage in poor condition is not a maintenance issue. It is a statement that the occupant did not choose to make and cannot withdraw. And it is made to the street, continuously, at all hours, to everyone.

We would like it on the record that the apology is never necessary and that we have never once thought less of anybody for the condition of their house. Most of the time the reason is straightforward. Nobody has been well, or money has been tight, or the person who used to do it is not there any more, or the house is simply tall and nobody has a ladder. None of that is a moral fact.

Section Five left something deliberately unfinished. The finding was Cutright's, reported in Mandel and colleagues' review. People who perceived a lack of personal control attempted to reassert it by choosing products that contained symbolic boundaries. The example being a framed painting rather than an unframed one.

A frame. A defined edge, placed around something, by somebody who is not in control of the wider situation.

Now look at what this trade actually does to a building.

Render cleaning. A rendered wall does not gradually darken all over. It grows algae and organic staining that creep inward from the edges. From under the eaves, up from the damp proof course, out from around the downpipe brackets, along the line where two elevations meet. What is lost first is the line between one surface and the next. What cleaning restores is not whiteness. It is the boundary between the wall and everything that is not the wall.

Block paving and driveway restoration. A drive does not vanish. Its margins vanish. Moss and growth come in from the kerb, from the border, from the join with the path, until the drive and the garden have no agreed frontier and the eye cannot find where one ends. Restoration is the recovery of the perimeter, and, in the joints, the recovery of several hundred small rectangles that had become one green mass.

Window cleaning. The centre of a pane rarely looks dirty. Light passes through it and the eye forgives a great deal. What is seen is the border. The algal line where the glass meets the frame, the run marks down the sides, the shadow at the bottom of the pane. A clean window is not a transparent surface. It is a defined rectangle.

Soffits, fascias and gutters. These are the boundary. They are the physical junction between the wall and the roof, and they are the trim that makes that junction read as deliberate. When they go, the house stops having a top edge.

Four services. Every one of them, described physically, is the restoration of an edge that had dissolved.

We now have to be careful, and this is the section where the temptation is strongest, because the fit is so neat that it is suspicious.

Cutright's experiment is about paintings in a laboratory. Not houses, not render, not paving, not anybody in Northamptonshire. There is no study showing that people who feel a loss of control book exterior cleaning, and we are not claiming there is. Altman and Low's collection is about attachment to place. It is not about buying services and it is not about us.

And there is an ordinary explanation available for everything above that requires no psychology whatsoever. Edges are where dirt collects. Water runs to edges. Organic growth starts where two materials meet, because that is where moisture sits. A cleaner restores edges because edges are what gets dirty, and if that is the whole story then this section is an elaborate way of describing drainage.

We think both are true, and we cannot prove the second one. The physical account is certainly correct. Whether the thing being bought, when a customer books a render clean before a family visit, is a physical improvement or a restored boundary, we do not know, and no amount of standing on a ladder will tell us.

What we can say is narrower and is not in dispute. The physical work of this trade and the thing Cutright measured have the same shape. We are placing them side by side. We are not welding them together.

There are three moments when this work gets booked at short notice and out of the usual rhythm, and they are worth listing because they are the same three every year.

Before a visit. Somebody is coming. Family, in laws, a christening, a wake. The house is about to be seen by people whose opinion matters, rather than by the street, which has stopped counting.

Before a sale. The frontage is about to be photographed and put in front of strangers who will form an entire view of the property from it. This is the most rational version of the three and nobody involved pretends otherwise.

After something has gone wrong. A period of illness, an absence, a bereavement, a stretch where the house was not the priority. The work that gets booked is not usually the most urgent thing on the building. It is the most visible thing.

We have no theory about that last one and we are not going to invent one. We are recording that it happens, because it is the single most consistent pattern in fourteen years of bookings and we have never seen it described anywhere.

So. Not clean glass, exactly. A defined rectangle where there was a smeared one. A line under the roof where the line had gone. A perimeter to the drive. An edge between the wall and the weather.

And, from Section Six, the recurrence of all of it, on a cycle, without having to ask.

Section Nine is about a system that proposes to represent all of this as a number between one and five.

EIGHT. THE MAN AT THE DOOR

Everything in this section comes from two sociologists, both long dead, neither of whom knew anything about window cleaning. Everett Hughes published Men and Their Work in 1958. Erving Goffman published Stigma in 1963. Between them they described what happens in the thirty seconds between a tradesman knocking and a householder deciding who he is, and they described it more precisely than anybody in the trade has managed since.

We are not complaining in this section. Hughes did not complain and neither will we. He was doing something more useful than complaining. He was working out how the machinery runs.

Hughes starts with a distinction that does most of the work in this section. This is from pages forty nine and fifty.

Now every occupation is not one but several activities. Some of them are the dirty work of that trade. It may be dirty in one of several ways. It may be simply physically disgusting. It may be a symbol of degradation, something that wounds one's dignity. Finally, it may be dirty work in that it in some way goes counter to the more heroic of our moral conceptions.

Three separate kinds, and they come apart. Something can be physically revolting and carry no shame at all. Something can be spotlessly clean and be humiliating. Something can be neither and still sit badly with the story a person tells about themselves.

Then the sentence that matters. Dirty work of some kind is found in all occupations. That is page fifty.

All of them. The solicitor has dirty work. The general practitioner has dirty work. The head teacher, the accountant, the estate agent, the person who signs off the redundancies. Every one of them does some part of the job they would rather not describe at a dinner party. Hughes goes further and says it is hard to imagine an occupation where a person is not, in certain repeated situations, practically compelled to play a role of which he thinks he ought to be a little ashamed.

So the difference between trades is not whether the dirty work exists. It is whether it can be kept out of sight. Of the janitor he was studying, Hughes says simply. This is patent. He could not hide it if he would.

That is the whole of it. We clear gutters. What comes out of a gutter after eighteen months is leaf mould, grit, moss, the roof's own eroded material, and, often enough, a dead bird. It goes into a bag, and the bag goes in the van, and anybody standing on the pavement can watch the entire process. There is no back office for a gutter. There is no discreet email in which the unpleasant part is handled by somebody more junior. The dirty work of this trade is performed at the front of the house, in daylight, in front of the street.

That is not a hardship. It is a methodological advantage, and it is why this document exists. But we will come back to that.

This next finding is counter intuitive enough that it is worth stating slowly. Hughes asked janitors what the worst part of the job was, and they were, in his word, bitterly frank about the physically dirty parts. But when he looked at where the injury actually sat, it was not in the material. It was in the relationship in which the material was handled. This is page fifty.

If it were not for a tenant who had broken a window, he could have got his regular Sunday cleaning done on time. If it were not for a tenant who had clogged a trap, he would not have been ignominiously called away from the head of his family table just when he was expansively offering his wife's critical relatives a second helping of pork chops, talking the while about the importance of his job. It is the tenant who causes the janitor's status pain.

Read what that passage is actually describing. A man at his own table. His wife's relatives, who are critical of him, are present. He is doing the thing every self employed person does in that situation, talking up the job, making the case for himself, and the job reaches into the room and takes him out of it.

The indignity is not the trap. The indignity is being removed from your own dinner, in front of the people whose opinion of you you were in the middle of managing.

We have no better name for this than the one we will give it ourselves, so we will say plainly that it is ours and not Hughes's. Call it the phenomenology of interruption. Hughes identified the mechanism. He did not name it. What he saw is that the wound is not administered by the work. It is administered by the timing of the work, and by who is watching when it lands.

Fourteen years in, we can confirm the finding from the other end of the century. The gutter that overflows does not overflow at eleven o'clock on a Tuesday. It overflows during a storm, on a Sunday, and the call comes at seven in the evening. The job itself is twenty minutes on a ladder. The cost is the evening.

That call gets answered by me, because every call to this business has always been answered by me. There is no rota and there never was one. We will come back to what that costs in Section Ten.

Hughes had no term for the customer in that scene, and the trade's own vocabulary for them is unprintable. There is, it turns out, an academic one.

In 1994 Christopher Lovelock coined the word jaycustomer, built the way jaywalker is built, for customers who, as Harris and Reynolds of Cardiff University summarise it, deliberately or unintentionally disrupt service in a manner that negatively affects the organisation or other customers.

We are printing that term for one reason. It lets us describe a behaviour rather than a person, and the distinction matters to us more than it might appear to.

Note the phrase, or unintentionally. Lovelock's category is not a category of bad people. Most disruption is not malicious and a great deal of it is not even noticed by the person doing it. The customer who rings at seven on a Sunday is, in their own mind, being reasonable. Something is overflowing and we are the people who deal with that. They are right about the second part.

What we will not do, here or anywhere in these six articles, is illustrate this from the round. No customer of ours appears in this document. Not named, not anonymised, not composited, not disguised by changing the street. That rule is absolute and it is the reason this section leans on 1958 Chicago instead of 2026 Northamptonshire. Hughes supplies the evidence so that we never have to.

Hughes then records something about his janitors that we are going to handle carefully, because it is the sharpest thing in the chapter and it is easy to misuse.

He found that the janitor learned a great deal about the tenants from their rubbish. Enough to form a detailed and unflattering picture of people who treated him as beneath their notice. And he found that the janitors did not act on it. This is page fifty one.

The garbage gives the janitor the makings of a kind of magical power over that pretentious villain, the tenant. I say a kind of magical power, for there appears to be no thought of betraying any individual and thus turning this knowledge into overt power. He protects the tenant, but, at least among Chicago janitors, it is not a loving protection.

It is not a loving protection. Six words, and they describe an arrangement that exists in every trade that works inside or immediately outside other people's homes, and that nobody has ever put on a website.

The discretion is real. It is also unenforced, unpaid and unacknowledged. There is no professional body, no confidentiality clause, no register. It holds because the people doing it have decided it should hold, and for no other reason.

We will say only this about our own version of it, and then stop. A window cleaner sees into rooms. That is not a hazard of the job, it is the job. The glass is the thing we are looking through. Six weeks after six weeks, for years, across four hundred houses. What is seen is not discussed. Not at the next house, not in the van, not at home, and not here.

Hughes noticed something else about those interviews, which is that his janitors spoke freely to him and could not have spoken that freely anywhere else. To do so, except in the most tightly closed circles, would create impossible situations. He was right, and it is why this paragraph ends here.

Hughes gives the trade two pieces of equipment. Both need handling honestly, because of what he made them out of.

The first is auxiliary characteristics. From page one hundred and three. There tends to grow up about a status, in addition to its specifically determining traits, a complex of auxiliary characteristics which come to be expected of its incumbents.

A status has a core, what the person actually does and is competent at, and it accumulates a cloud of expected extras that have nothing to do with the competence and are treated as though they did. What sort of person does this. How they speak. What they read. Whether they went anywhere after school.

The second is the master status determining trait. A trait that, in his words on page one hundred and eleven, tends to overpower, in most crucial situations, any other characteristics which might run counter to it.

And here is the honest part, which is not optional. Hughes forged that concept in 1945 to describe American racism. The paper is Dilemmas and Contradictions of Status, published in the American Journal of Sociology in March of that year, and the case he is working through is what happened to a Black American who qualified as a doctor or a lawyer. The profession being real, and being overpowered anyway.

We are borrowing the tool and leaving the case where it belongs. Manual trade is not race and this article does not suggest for a moment that it is. What Hughes demonstrated is that some traits are structurally louder than others, and that competence does not automatically win the argument. That mechanism generalises. The injury he was describing does not, and we are not going to pretend otherwise by quoting him without saying what he was quoting about.

This next one made us sit up, because it explains something about this trade that we had felt for years and never had a sentence for. Hughes is working out how people manage the awkwardness of a status that does not sit neatly, and he finds that the main technique is architectural. From page one hundred and twelve.

One way of reducing status conflict is to keep the relationship formal and specific. This is best done by walking through a door into a place designed for the specific relationship, a door which can be firmly closed when one leaves.

The consulting room. The office. The interview room. The counter. Whatever passes between two people inside it is bounded by the room, and the door ends it.

And then, at the other extreme, on the same page. At the other extreme is school teaching, which is done in full view of the community and is generally expected to be accompanied by an active role in community activities.

A trade that works on the outside of buildings has no door at all.

Every part of this job is done in public. The ladder goes up in the street. The van is parked where the neighbours can read it. The conversation about the price happens on a doorstep, at normal speaking volume, with somebody's front garden as the only acoustic barrier. If a job goes wrong it goes wrong where people can see. If a customer is difficult, they are difficult outdoors.

And, the part Hughes's teachers would recognise instantly, we live here. Hartwell is a village. The relationship does not end at the boundary of a room, because there is no room, and because the person whose gutters we cleared on Tuesday is in the same shop on Saturday.

That is not a grievance. It is, in fact, most of why the work is worth doing. A trade with no door is a trade that cannot quietly get away with anything, and over fourteen years that has been a discipline rather than a burden. But it is a structural fact about the job, Hughes identified it in 1958, and it explains why a bad thirty seconds on a doorstep costs more here than the same thirty seconds costs in an office.

Goffman takes over here, and his contribution is the sharpest description we have found of the first moments of a job.

He observes that we walk up to a stranger already carrying a set of assumptions about what they ought to be. Assumptions we did not notice ourselves making, and only notice at all when they fail. He calls the person we have imagined the virtual social identity, and the person who can in fact be proved to exist the actual social identity. This is page two of Stigma.

The character we impute to the individual might better be seen as an imputation made in potential retrospect, a characterisation in effect, a virtual social identity. The category and attributes he could in fact be proved to possess will be called his actual social identity.

The gap between those two is where the entire doorstep transaction happens.

And Goffman is immediately careful about something that saves this section from becoming a complaint. This is page three.

The term stigma, then, will be used to refer to an attribute that is deeply discrediting, but it should be seen that a language of relationships, not attributes, is really needed. An attribute that stigmatises one type of possessor can confirm the usualness of another, and therefore is neither creditable nor discreditable as a thing in itself.

Nothing about cleaning windows is discrediting in itself. Goffman is explicit and he is right. What does the work is the relationship between the attribute and the expectation. And he proves it with an example that is worth repeating exactly, from the same page. For example, some jobs in America cause holders without the expected college education to conceal this fact. Other jobs, however, can lead the few of their holders who have a higher education to keep this a secret.

The same fact. Concealed in both directions. That is the whole mechanism, and it is why this section is not a complaint about anybody's manners.

Goffman then describes something we had assumed was ours alone. This is page fourteen.

During mixed contacts, the stigmatised individual is likely to feel that he is on, having to be self conscious and calculating about the impression he is making, to a degree and in areas of conduct which he assumes others are not. Also, he is likely to feel that the usual scheme of interpretation for everyday events has been undermined. His minor accomplishments, he feels, may be assessed as signs of remarkable and noteworthy capacities in the circumstances.

In the circumstances. That is the phrase.

His first illustration is a professional criminal being told by a visitor, in apparent delight, that it is amazing he should read books like this, Cockburn, Klare, de Beauvoir, Durrell, rather than paperback thrillers with lurid covers.

His second is a blind man, from pages fourteen and fifteen.

His once most ordinary deeds, walking nonchalantly up the street, locating the peas on his plate, lighting a cigarette, are no longer ordinary. He becomes an unusual person. If he performs them with finesse and assurance they excite the same kind of wonderment inspired by a magician who pulls rabbits out of hats.

Here is the reason we can print this without accusing a single person of anything. In Goffman's material, the people delivering these remarks are being kind. They mean it warmly. They would be horrified to learn how it landed. The criminal in the example knows perfectly well that the speaker meant no harm, and is furious anyway, and both of those things are true at once.

We have had, over fourteen years, a number of customers express surprise that we write things down. That we photograph the work. That there is a method. That somebody who cleans render has read anything at all. Every one of those remarks was meant as a compliment and every one of them was a compliment, and every one of them also contained the words, in the circumstances, unspoken, doing the work.

We are not asking anybody to stop. We are noting, with Goffman's help, what the sentence is made of.

Goffman's other major distinction is between the discredited, where the thing is already visible and known, and the discreditable, where it is not yet known and could be managed, concealed, or revealed.

He treats those as two different predicaments requiring two different strategies. A tradesman on a doorstep has both simultaneously, and that is our inference from his framework rather than anything he wrote.

The manual work is discredited. It arrived before we did. The van said it, the ladder said it, the overalls said it. There is nothing to manage because there is nothing left to disclose.

Whatever else a person has, reading, training, a mind, a second life, anything that sits outside the auxiliary characteristics expected of the status, is discreditable. Invisible, therefore a live decision. Do you mention it? If you do, you are either showing off or explaining yourself, and both readings are available to the listener. If you do not, you have agreed to the imputation.

Two opposite problems, in the same conversation, on the same doorstep, in about thirty seconds. Goffman gave us the two concepts. Putting them on top of each other is ours.

Goffman divides stigma into three types. Bodily, character, and what he calls tribal. And in a footnote to that passage, writing in 1963, he adds this. In recent history, especially in Britain, low class status functioned as an important tribal stigma, the sins of the parents, or at least their milieu, being visited on the child.

We are going to let him say that and add nothing to it. It is a sentence written by an American sociologist about Britain, in a footnote, in 1963, in a book about something else entirely. It is better evidence for being incidental, and it is better coming from him than from us.

There is a second Goffman finding, that a single failing tends to be read as typical of the whole category rather than as an event, which is central to how reviewing works and which we have already dealt with at length. It is Article Three's, not this one's.

Hughes ends the chapter with the finding that turned out to be the most useful thing in either book, and it is the reason Section Twelve exists.

He notices that some high prestige occupations cannot delegate their dirty work at all. The physician handles the human body, and that handling is the thing that gives the profession its standing. So the unpleasant part is not hived off. It is absorbed. This is page fifty two.

In this case, I suppose the dirty work is somehow integrated into the whole, and into the prestige bearing role of the person who does it. The janitor, however, does not integrate his dirty work into any deeply satisfying definition of his role that might liquidate his antagonism to the people whose dirt he handles.

And then he proposes the classification. We might conceive of a classification of occupations involving dirty work into those in which it is knit into some satisfying and prestige giving definition of role, and those in which it is not.

Read that twice, because it is not a description of an injustice. It is a specification.

Hughes is not saying the physician deserves more and the janitor deserves less. He is saying the physician's trade has built a definition of its own role large enough to contain the worst part of the work, and the janitor's trade has not built one. The dirt is identical. The container is not.

That is a gap in the equipment, and gaps in equipment can be closed. What we did about it is Section Twelve.

NINE. RENTING ACCESS TO YOUR OWN NEIGHBOURS

There is a proposition put to every small trade in Britain, and it is put so often and so confidently that it has stopped sounding like a proposition at all. It sounds like weather. The customers are online now. You have to be where the customers are. You pay for access, and access is what you get.

We want to take that apart. Not because we are against it, we have paid for leads and we will say what that cost, but because the sentence contains four claims dressed as one, and nobody in this trade has ever been given the vocabulary to separate them. The people who have that vocabulary are economists and sociologists writing about something else entirely. That is usually how it goes.

Start with the definition, because the definition is doing more work than it appears to. Nick Srnicek, in Platform Capitalism, published by Polity in 2017, page forty three. At the most general level, platforms are digital infrastructures that enable two or more groups to interact.

Twelve words. Read them again and supply the two groups yourself.

For a lead generation service in the exterior cleaning trade, group one is householders who want a job done. Group two is us. The platform is the infrastructure in between. It does not clean anything. It is not a cleaning company that has grown large. It is a different kind of firm altogether, and the confusion on that point is the source of most of the bad advice in this sector.

Srnicek sorts these firms into five kinds. Four of them are not us. Advertising platforms, cloud platforms, industrial platforms, product platforms. The fifth, from pages forty nine and fifty, is the lean platform. Those which attempt to reduce their ownership of assets to a minimum and to profit by reducing costs as much as possible.

Hold that definition against this business and check it item by item, because it fits without a single adjustment.

We own two ladders and a pole system. We own a van and we pay to insure it, tax it, put it through its MOT and fuel it. We own a water tank, a filtration setup and a pressure washer. We carry public liability cover. We hold a waste carrier registration for what comes out of the gutters. If a pane gets scratched, we pay for the pane. If a lightwell floods, we deal with the lightwell. If somebody falls off a ladder, it is one of us.

A lead platform owns none of that and risks none of it. It owns the infrastructure in between, and it profits from the introduction.

That is not an accusation. Srnicek is describing a business model, not a crime, and it is a perfectly coherent business model. But it means the relationship is not what it is usually described as. We are not a supplier to the platform and it is not a customer of ours. It is a landlord of a position, and we rent the position.

Here is the finding that reorganised this whole section for us, and it comes from a sentence Srnicek writes almost in passing. He is explaining why lean platforms classify the people who do the work as contractors rather than employees. This is pages seventy six and seventy seven.

In America, these platforms legally understand their workers as independent contractors rather than employees. This enables the companies to save around thirty per cent on labour costs by cutting out benefits, overtime, sick days, and other costs. It also means outsourcing training costs, since training is only permitted for employees. And this process has led to alternative forms of control via reputation systems, which often transmit the gendered and racist biases of society. Contractors are then paid by the task.

We are quoting Srnicek's sentence whole, including his clause about bias, because stopping a quotation just before an author's uncomfortable point is a way of editing him without saying so. That clause is his claim, not ours, it is a serious one, and a reader who goes to the book should find nothing we quietly removed.

Now take the middle of it on its own. Reputation systems are an alternative form of control.

Not a consumer information service. Not a quality mechanism. Not a neutral record of how a job went. A substitute for management, adopted precisely because the platform has declined the employment relationship and still needs a way to direct behaviour.

Think about what an employer normally does. Trains you. Supervises you. Tells you when you have got it wrong. Carries the cost of all three, and carries the liability when it goes wrong anyway. A platform that classifies you as a contractor has shed every one of those costs, and has replaced the entire function with a number administered by strangers.

And it is a very efficient substitute, because it costs the platform nothing. The customers do the supervising for free, and the person supervised absorbs every consequence.

We will put this as plainly as we can, because it took us years to see it. The star rating is not there to tell customers about tradesmen. It is there to do the job a foreman used to do, without anybody having to employ a foreman.

Which lands exactly where Section One and Section Three landed. Porter's finding was that quantification substitutes for judgement where there is no trusted community to do the judging. Srnicek's is that it also substitutes for supervision where there is no employer to do the supervising. The number replaced the neighbour's word and the foreman's eye at the same time, and the person being counted is paying for both.

The obvious response is, fine, then do not use it. Cory Doctorow explains why that is harder than it sounds, and he does it with three pieces of ordinary economics that have nothing to do with anybody's opinion of anybody.

The first is network effects. A service gets more valuable as more people use it. In his opening chapter. You joined Facebook because the people who were already there made it valuable to you, and once you were there, you made Facebook more valuable to the people who wanted to hang out with you.

Change the nouns. You join a lead platform because the customers are there. And by joining, you become part of the reason customers go there, because a platform with plenty of tradesmen on it is a platform worth using.

The second is switching costs. Switching costs are everything you have to give up when you switch from one product or service to another.

For a tradesman, the switching cost has a specific name. The review history. Two years of work, several hundred jobs, every one of them rated. It is the most valuable thing a small operator accumulates on a platform, and it is the one thing that cannot leave with him.

The third is the collective action problem. In his words, the collective action problem is the incredibly difficult business of getting other people to do what you want them to do, when you want them to do it. And, adding people to the group makes the collective action problem exponentially more problematic.

And then the sentence that does the work. So Facebook users cannot help but take one another hostage, and Facebook grew progressively more cognisant of this fact.

Substitute tradesmen on a lead platform for Facebook users and nothing else in that sentence needs to change.

No single window cleaner can leave, because leaving means abandoning the place the customers are. The customers are there because the tradesmen are. Every contractor on the platform is part of the reason every other contractor cannot leave. Nobody intends this. Nobody could unilaterally fix it. It is not a conspiracy. It is arithmetic.

There is an obvious question sitting underneath all of that, and it took us three drafts to notice we had never answered it. Why can you not take the reviews with you?

There is no technical reason. A review is a line of text, a number and a date. Exporting several hundred of them is a trivial engineering problem, vastly easier than the thing the platform already does every day. Bank accounts are portable. Mobile numbers are portable. Pension pots are portable. Reputations are not.

Srnicek answers it in his conclusion, on page one hundred and thirteen, and the answer is not sinister, which is somehow worse.

The tendency of major platforms to grow to immense size thanks to network effects, combined with the tendency to converge towards a similar form, as market pressures dictate, leads them to use enclosure as a key means of competing against their rivals. If this analysis is right, then capitalist competition is driving the internet to fragment.

The wall is not an oversight. The wall is the competitive strategy.

A platform that let you export your reputation would be handing its rivals its own moat. It would be the single most self destructive thing it could build. So nobody built it, and nobody is going to, and no amount of complaining will produce it, because the absence is not a bug that somebody forgot to fix.

Two writers reach that finding independently. Srnicek from an academic monograph, Doctorow from a trade book for general readers. When two people arrive at the same mechanism from opposite ends of the shelf, it is worth more than either of them saying it twice.

We should say what this looks like from a ladder, because none of the above is abstract when you are standing on one.

A job comes in. Four other firms get the same job at the same moment. All five of us price it. One of us gets it, which means four of us have spent the morning pricing work we will not do, and every one of those four has paid for the privilege of being told about it.

Sometimes the address is a street we already work. That is the part that takes some getting used to. Paying a company in another county for an introduction to a house forty yards from a house we have cleaned every six weeks for nine years.

We name no platform in this article. Not one. Article Four set that rule and it holds here. We describe mechanisms, and we name a company only where there is documented conduct to attach to the name. We have none. What we have is a structure, and the structure is the point.

Everything above is about the terms of the deal. This next part is about whether the thing being sold can be sold at all.

In April 1997, before Google, before Facebook, before any of it, a former theoretical physicist named Michael Goldhaber gave a paper at Harvard's Kennedy School and published it in First Monday. He argued that the internet would not produce an information economy, because information was not scarce. It would produce an attention economy, because attention is. Attention, at least the kind we care about, he writes, is an intrinsically scarce resource.

He then describes exactly what a performer is selling, and it is not information.

What I would suggest is going on is less that I am providing you with information that you deem in advance will be of value, than that I am offering you individually the illusion of my full attention.

And, in the same section, that anyone who speaks or writes or seeks attention in any way has to become something of a success in the special rhetoric of persuading listeners, readers, and so on, that he or she is meeting their individual needs, when in fact some of these needs have been artfully set up in advance.

And then the finding this section has been walking towards. Money flows to attention, and much less well does attention flow to money.

He puts it even more bluntly a few lines earlier. Contrary to what you are sometimes urged to believe, money cannot reliably buy attention.

That is the whole, you have to pay to be where the customers are, proposition, answered in 1997 by a man who had never heard of it. The flow runs one way. A business that already has attention can convert it into money without much difficulty. A business with money cannot reliably convert it back.

And Goldhaber adds one more thing that reads, in 2026, like a description of a review history. In the new economy attention itself is property. Where is it? Primarily it is located in the minds of those who have paid you attention in the past, whether years ago or seconds ago.

Property located in other people's minds. Which is precisely what a round is, four hundred households who know who we are and expect us on a particular Thursday, and precisely what a platform review history is not, because a review history is property located in somebody else's database.

The round cannot be confiscated. It can only be earned and lost. The rating sits on a server owned by a third party, and the arrangement under which we hold it can be varied by that third party at any time, without our agreement.

One last Goldhaber, because it explains Section Ten before we get there. Since it is hard to get new attention by repeating exactly what you or someone else has done before, this new economy is based on endless originality, or at least attempts at originality.

Endless. Not a campaign with an end. An appetite.

We want to close this section in 1900, with a German sociologist who never saw a telephone directory, let alone a comparison engine.

Georg Simmel, in The Philosophy of Money, is describing what happens to people at the top of a money culture. Part of his argument is this. The effort of acquiring something becomes part of the thing acquired. The detours, the difficulty, the specific demands a thing makes on you before you can have it, all of that transfers into the object and is afterwards experienced as a quality of the object. Some of a thing's interest is made of what it cost you to get it.

From which it follows that when acquisition becomes mechanical and indifferent, the thing acquired becomes colourless. Not because of what it is. Because of how it arrived. This is page two hundred and fifty seven, in the Bottomore and Frisby translation.

The more the acquisition is carried out in a mechanical and indifferent way, the more the object appears to be colourless and without interest, just as everywhere it is the case that not only does the goal determine the way, but also the way determines the goal.

Not only does the goal determine the way, but also the way determines the goal.

Consider two routes to the identical job, done by the identical people, to the identical standard.

Route one. A neighbour mentions us over a fence. The customer waits until we are next on the street. There is a conversation at a door. A specific arrangement is made with a specific person. Getting there has cost the customer something. Attention, a small act of trust, the mild effort of asking somebody else what they thought.

Route two. A form is completed. Four quotes arrive within the hour. The lowest is accepted. Procurement has been equalised. No particular method has been reserved for this purchase rather than any other.

Simmel's claim is that these two routes do not deliver the same object. The second has been acquired by an indifferent method, and the indifference of the method transfers into the thing.

We need to be careful here and we are going to be explicit about it. There is no study showing that customers value platform sourced work less than referral sourced work. We are not claiming there is. What we are doing is placing a philosopher's account of how the manner of acquisition enters the thing acquired beside fourteen years of watching two channels behave differently, and saying, these appear to be about the same thing. We are not saying the first proves the second.

Nor is this nostalgia, and it must not be read as nostalgia. Simmel is not saying the old way was nicer. He is making a claim about how value is constituted, and he is making it in 1900, about money, at a moment when the thing he was describing was brand new.

What it gives this trade is a sentence for something everybody in it knows from the inside and has never had words for. Work sourced through a platform is harder to be valued for, and the difficulty is built into the channel rather than into the customer.

The platform's entire selling proposition is the removal of the customer's effort. Simmel identified that effort, in 1900, as the thing that deposits value into the object before the work has even started.

Six things this section is not saying, stated flatly, because a section like this collects misreadings.

One. We are not saying platforms are fraudulent. Nothing here alleges dishonesty by anyone. Every mechanism described is ordinary competition behaving exactly as economics predicts.

Two. We are not saying do not use them. We have used them. Plenty of good firms use them well.

Three. We are not saying customers are wrong to use them. A householder with a leaking gutter and no neighbour to ask is behaving entirely sensibly.

Four. We are not naming anyone, for the reason given above.

Five. We are not claiming a proof. Simmel is a frame. Doctorow is a trade book. Srnicek and Goldhaber are the evidence, and we have said which is which.

Six. We are not complaining. We are describing a structure we operate inside, and we would rather understand it than resent it.

What we are saying is narrower and, we think, more useful. The thing being sold is access to attention. The thing that accumulates is property held in somebody else's system. And the reason it cannot be moved is that immobility is what the seller is competing on. Once that is clear, the decision about whether to buy becomes a real decision instead of an article of faith.

TEN. THE PERFORMANCE AFTER THE SHIFT

Section Nine was about what a platform charges. This one is about what it asks for that has no price on it at all.

The proposition is familiar to every small trade in the country. Post your work. Show your face. Film the before and after. Be relatable. Build an audience. And, the word that does most of the work, be authentic, which in this context means, do all of the above while appearing not to be doing any of it.

We want to describe precisely what is being asked for there, because there is a name for it, it was named in 1983, and once you have the name the request looks different.

Arlie Russell Hochschild's The Managed Heart is a study of flight attendants and bill collectors, published in 1983. Her subject is what happens when feeling itself becomes part of the job. Her definition sits in a footnote on page seven, and it is the most useful footnote in this article.

I use the term emotional labour to mean the management of feeling to create a publicly observable facial and bodily display. Emotional labour is sold for a wage and therefore has exchange value. I use the synonymous terms emotion work or emotion management to refer to these same acts done in a private context where they have use value.

The same acts. Two different values, depending on who they are for.

In the body of the page she describes what the labour consists of. This labour requires one to induce or suppress feeling in order to sustain the outward countenance that produces the proper state of mind in others. This kind of labour calls for a coordination of mind and feeling, and it sometimes draws on a source of self that we honour as deep and integral to our individuality.

And then the sentence that we cannot improve on. The company lays claim not simply to her physical motions, how she handles food trays, but to her emotional actions, and the way they show in the ease of a smile.

The flight attendants she interviewed had their own phrase for the result. They spoke of their smiles as being on them but not of them. An extension of the make up, the uniform, the recorded music and the pastel decor, all of it orchestrating the mood of the passengers.

Now apply it to a doorstep.

A tradesman being pleasant at a door in 2004 was doing emotion work. He was managing his manner, adjusting his tone, being patient with somebody who was anxious about the price. It had use value. That is simply how you behave towards a neighbour, and it belonged to him.

The same man, in 2026, filming a render clean for an audience, writing a caption in a voice that is recognisably his, being relatable on a schedule, is doing emotional labour. It is managed feeling produced for public display, in a commercial context, to a standard set elsewhere. Except that nobody is paying him a wage for it.

Hochschild's flight attendant was employed. Her emotional labour was extracted by a company that trained her, supervised her, and put her on a payroll. However badly that bargain went, and Hochschild documents in detail how badly, it was a bargain.

The company laying claim to a tradesman's emotional actions is a platform he does not work for, has no contract with, cannot invoice, and cannot appeal to.

The extraction survived. The employment relationship did not.

That is the same finding Section Nine reached from the other end. Srnicek's observation that reputation systems became a form of control precisely because the employment relationship was declined. Here it is again, in a different currency. The rating replaced the supervisor. The content replaced the advertising budget. Neither replacement came with the thing the original had attached to it.

Brooke Erin Duffy studies people who produce content for a living, or hope to. Her term for what they are doing is precise. This is page eleven of the publisher preview of her book, published by Yale University Press.

Aspirational labour is a mode of mostly uncompensated, independent work that is propelled by the much venerated ideal of getting paid to do what you love. As both a practice and a worker ideology, aspirational labour shifts content creators' focus from the present to the future, dangling the prospect of a career where labour and leisure coexist.

Mostly uncompensated. Propelled by an ideal. Focus shifted from the present to the future.

Duffy reports one figure about what the future actually pays, and we are going to handle it exactly as Section One requires. A survey by Blogging dot com of one thousand bloggers found that seventeen per cent are able to sustain their lifestyle or support their family with their blogs, while eighty one per cent never make even one hundred dollars from blogging.

Now the health warning, in the same breath as the number, because that is the rule.

Blogging dot com is a commercial operator in the market it surveyed. The sample was self selected. The figure is not peer reviewed. Duffy herself hedges in the very next sentence, noting that the statistics describe blogging activities more generally. It is a marketing survey, published by an interested party, reported in a Yale University Press book.

We are printing it anyway, with all of that attached, because it is the only figure available and because burying its provenance while quoting it would be precisely the behaviour Section One condemns. A university imprint does not launder a vendor's survey, and an article about the parentage of numbers does not get an exemption for a number it likes.

And a limit on Duffy herself. We hold a thirty two page publisher preview, not the book. Nothing beyond those pages is attributed to her, here or anywhere.

Here is the part of this section that changed how we understand our own inbox.

Horton and Wohl, in 1956, described what accumulates when somebody appears regularly. The persona offers, above all, a continuing relationship. His appearance is a regular and dependable event, to be counted on, planned for, and integrated into the routines of daily life. Their continued association with him acquires a history, and the accumulation of shared past experiences gives additional meaning to the present performance.

Section Six used that passage to describe a round, and it fitted. Now here is the sentence Section Six deliberately withheld.

The crucial difference in experience obviously lies in the lack of effective reciprocity, and this the audience cannot normally conceal from itself.

Effective reciprocity. That is the term this article has been circling since Section Six, and it is the exact difference between a round and a feed.

On a round, reciprocity is effective. We turn up, the gate is unlocked. We do the work, the money is under the mat. There is a conversation at a door and both people are in it. The history accumulates on both sides. We know things about them and they know things about us, and neither party is guessing.

In a feed, it is not. The history accumulates on one side only. Somebody has watched forty videos of a render clean. They know the van, the voice, the sequence, the way the wall comes back. We know nothing whatsoever about them. They have a relationship. We have an audience metric.

And Horton and Wohl's clause is the humane part. The audience cannot normally conceal the asymmetry from itself. They know, at some level, that it is one directional. They feel it, and the feeling is uncomfortable, and one of the things people do about it is close the gap themselves.

Which is why a stranger sends a message at half past ten on a Sunday night and is genuinely surprised to get no answer.

They are not being rude by their own lights. They think they know you. And on the evidence of everything they have seen, that belief is not unreasonable. It is just structurally impossible to correct, because the thing that produced it is designed to produce it.

Horton and Wohl also describe the far end of this, where the follower demands real reciprocity which the para social relation cannot provide. We are not going to develop that, because it goes somewhere this article does not need to go, and because it is a description of distress rather than of rudeness.

There is one more mechanism, and it needs handling carefully because it is routinely misused.

John Suler, writing in CyberPsychology and Behaviour in 2004, set out six factors that produce what he called the online disinhibition effect. The reason people say and do things online that they would not do face to face. His factors are dissociative anonymity, invisibility, asynchronicity, solipsistic introjection, dissociative imagination, and minimisation of authority.

And then he says the thing that most people who cite him leave out. This is page three hundred and twenty one.

Rather than thinking of disinhibition as the revealing of an underlying true self, we can conceptualise it as a shift to a constellation within self structure, involving clusters of affect and cognition that differ from the in person constellation.

Not a true self revealed. A different configuration, produced by a different architecture.

This is the single most important sentence in this section, and it is the one that stops the section becoming a complaint about people.

The popular version of Suler, that the internet reveals what people are really like, is a claim he explicitly declines to make. The architecture changed what they did. Anonymity, invisibility and asynchronicity are properties of a system, not properties of a person, and a person who is short with you in a comment and perfectly decent at a door is not being two faced. They are in two different rooms, and the rooms are built differently.

We will not use this to characterise any individual, and there is no example here. Suler's finding is about architecture. It is deployed here against the system, and it would be a straightforward abuse of it to deploy it against a person.

Section Nine left Goldhaber's observation lying on the table. Here is what it is for. Since it is hard to get new attention by repeating exactly what you or someone else has done before, this new economy is based on endless originality, or at least attempts at originality.

Endless. Not a campaign with an end date. Not a body of work that can be completed and then drawn on. An appetite that resets.

And Simmel, in 1900, describes the appetite from the demand side. The craving that a money culture produces for stimulation as such. On page two hundred and fifty eight he describes the modern preference for stimulation as such in impressions, relations and information, without thinking it important for us to find out why these stimulate us.

Without thinking it important to find out why.

Put those two together and you have the reason the performance cannot be completed and banked. The audience is not looking for information about window cleaning. It is not looking for anything in particular. It is looking for stimulation, and stimulation does not accumulate. Each instance is spent on arrival.

Which means there is no quantity of it that constitutes enough. A round can be complete. Four hundred houses, a six week cycle, a full week's work. A feed cannot be complete. It can only be current, and currency expires daily.

Simmel is a philosopher writing in 1900 and this is his account of a disposition, not a measurement of anybody's behaviour. We are placing it beside Goldhaber and Hochschild because three people, in three disciplines, across 1900, 1983 and 1997, arrived at the same shape from different directions. Independent agreement across that distance is worth more than any one of them alone, and it is still not proof.

One note on scope. This section is about performative labour only. The administrative burden, the quoting software, the invoicing, the compliance, the record keeping, the sheer volume of screen time a one van business now carries, is Article Five's subject and is not repeated here. It is real, it is separate, and it is dealt with there.

What this looks like at seven in the evening. The work ends and the other work starts.

Messages arrive in the evening because that is when people are at home thinking about their houses. That is not unreasonable of them. It is when the problem is in front of them. A question that takes ninety seconds to answer takes rather more than ninety seconds when there are eleven of them and they arrive between six and ten.

All of it is answered by one person, and always has been. There is no division of this work and there never was. That is not a complaint. It is a one man business and this is what a one man business is. But it means the cost does not get shared out, and it does not appear in any account of what this business does.

What has changed is the expectation. We used to try to answer within twenty four hours. We now answer when we answer. Deliberately, to take the pressure off a standard set by companies that deliver parcels. We are human and busy. We are not a dispatch system.

And there is a real difference between a customer and an audience, which took us a long time to see.

A customer has a house, an address, a problem and a price. The exchange is complete when the work is done and paid for. Effective reciprocity, in Horton and Wohl's phrase. And then it ends, cleanly, until the next cycle.

An audience has none of those things, is not owed anything, and cannot be satisfied, because it did not come with a specific want that could be met. It came for stimulation as such.

The trade is currently being advised to convert the first into the second, and to regard that as growth.

ELEVEN. THE PATTERN UNDERNEATH

We asked the person who runs this business to name, in one sentence, the shape he had been describing across all six of these articles. His answer. Life. Everything. For us, for them, and for the industry.

That is a bigger claim than we expected and it is the right one, so this section is built to it rather than to something narrower that would have been easier to write.

The argument is that the substitution described in Section One is not something being done to small businesses. It is happening in three directions at once, to three different parties, and the fact that nobody can see the other two is part of how it works.

Section One established the mechanism and Section Three applied it, so one sentence here. Theodore Porter's finding is that quantification spreads not because numbers are more accurate but because they are reached for where trust has run out, and that a decision rendered as a measurement acquires the appearance of not having been anybody's decision. His word was impersonal. Not unbiased, but without a person in it.

Hold that and watch it happen three times.

One. To us.

Section Four described five years in which the public account of this business was substantially written by people who had never bought anything from it. What matters here is not the unfairness. It is the structure. The format that carries the judgement has no field for the circumstances of the work.

We do not work in the rain. That is not a preference. Glass cleaned in rain is glass cleaned badly, and working at height in wind is how people get hurt. There is no box for it.

A hospital appointment cannot be moved to suit a schedule. There is no box for that either.

And the person who does the work is disabled. That is stated once in Section Two and once here, as a fact about how this business operates, and it will not be mentioned again. There is no box for it anywhere.

None of those is a defence, because the format does not take defences. It takes a number and a paragraph, and the paragraph is not addressed to us. There is no room in the mechanism for the reason, which is the same as saying that the mechanism was built on the assumption that reasons do not exist.

That is Porter's impersonal, arriving at a specific address.

Two. To them.

This is the direction we had not considered before he named it, and it is the one that makes the section worth writing. The same substitution has happened to the customer, and it has cost them more than it has cost us.

Section Two describes what a householder in 2001 was judging. Their own windows. Then, once the work had been done, fair pricing, flexibility, regularity and quality. And one visible physical thing, the frames, which could be checked from a pavement against the house next door by anybody with eyes.

That is not a crude system. It is a skilled one. It required attention, memory, a view of one's own property, and a willingness to form a judgement and act on it.

Section Four describes what a householder in 2026 has instead. Customers have no idea what they are paying for other than pure water, and everyone now uses a brush. That is the common perception and the qualifying data for consumers.

Two words, both of which describe equipment, neither of which distinguishes anybody. Plus a star rating, written mostly by people whose relationship to the work is unknown.

The customer has been deskilled as a judge.

They were not asked whether they wanted this. Nobody offered them a choice between assessing a frontage themselves and delegating it to an average. The capacity simply stopped being exercised, because something arrived that appeared to do the job, and the thing that appeared to do the job is thinner than what it replaced.

We would put it more plainly, and we mark this sentence as ours. A person who can no longer tell a good job from a bad one has lost something, and the loss is not visible to them, because the number arrives looking like information.

Porter said quantification is a way of making decisions without seeming to decide. The customer is the one doing the deciding. The rating lets them not notice that they have stopped.

Three. To the trade.

And the same shape again, at the level of the industry, and here it is measurable in tools.

Section Four recorded it. The cost of doing this work has risen sharply, fuel above all, while the going rate has not moved. Twelve to fifteen pounds is still common. So rounds contract into tighter geography to strip out travel, and when equipment breaks it is not replaced, because the rate no longer generates the money to replace it.

The result is that operators who began with a full set of tools, gutter kit, pressure washer, render equipment, end up doing windows only. Not by choice. By attrition.

That account comes from trade forums, trade posts and twenty five years of conversations with other operators. It is not a survey and we are not presenting it as one.

But look at what it describes. A market in which the only visible qualifying signals are pure water and a brush is a market in which nothing except price can be compared. And a market in which nothing except price can be compared will drive the rate to a level that cannot sustain the equipment needed to offer anything else.

The trade did not narrow because the work got simpler. It narrowed because the vocabulary for describing the work got simpler, and the equipment followed the vocabulary down.

That is the same substitution, a third time, arriving as an economic outcome rather than as a judgement about anybody.

Three parties. One structure. Something has to decide about you. It has no relationship with you. So it substitutes a measurement for a judgement, and then presents the measurement as though nobody had decided anything.

A rating. A score. A threshold. A criterion. A pair of standardised words that everybody now says.

And the same two properties every time, which are the properties that make it attractive to whoever installs it. It is cheap, because the people supplying the measurement do it for nothing and the person measured absorbs the consequences. And it is unaccountable, because there is no one to appeal to. Porter's word again. Not unfair. Impersonal. An unfair decision has a person in it who could be argued with. This does not.

We are going to state the wider claim in one flat sentence and then stop, because the alternative is a list of grievances, and a list of grievances is the thing that lets a reader dismiss all six of these articles.

The person who runs this business has encountered the same structure repeatedly. In commercial settings, in institutional ones, and in circumstances that have nothing to do with window cleaning. A decision made about him by something that had no relationship with him, on a criterion, with no route to explain anything, and no person in it to explain it to.

That is the sentence. It is placed once and we are not returning to it.

Article Two deals with one instance of it in detail, in the financial setting where it is best documented. This article does not reprint it, and the reason we are being this brief is not discretion. It is that the sentence is stronger than the stories. The moment it becomes a catalogue, a reader starts assessing each item on its merits and stops seeing the shape, and the shape is the entire finding.

Three things this is not.

It is not a conspiracy. Nobody convened. The people who built rating systems were solving a real problem, how do you decide about somebody you cannot know, and rating systems are a reasonable answer to that question. Section Nine went to some length to show that every mechanism it described was ordinary competition behaving exactly as economics predicts.

It is not new. Porter published in 1995 about developments running back through the nineteenth century. What is new is the reach. A mechanism that once applied to insurance underwriting and civil service examinations now applies to whether a window cleaner gets a phone call.

And it is not a claim that measurement is bad. Section One is built on an Office for National Statistics series and would be worthless without it. A figure with a parent, a method and a published series identifier is one of the most useful things a person can be handed.

The claim is narrower. That reaching for a number is evidence about the state of trust in a situation, and that where the number arrives, something else leaves. Sometimes what leaves is worth less than what arrives. Often it is not. Almost always, nobody checks which, because checking would require the judgement that the number was brought in to replace.

There is no answer to this at the scale of a one van business, and Section Fifteen will not pretend otherwise. But there is something, and it is what Section Twelve is about.

If a trade has no account of itself large enough to hold the difficult parts of the work, then any passing system will supply one, and the one that gets supplied will be a number. Not because the number is good. Because it was the only thing there.

TWELVE. WHAT THIS BUSINESS ACTUALLY DOES

Section Eight ended on a finding by Everett Hughes that we deliberately left standing, because it is the one thing in this entire article that could be acted on.

Hughes noticed that some high prestige trades cannot delegate their dirty work at all. The physician handles the human body, and that handling is the thing that gives the profession its standing, so the unpleasant part cannot be hived off to somebody junior. It has to be absorbed. The janitor, he writes on page fifty two, does not integrate his dirty work into any deeply satisfying definition of his role that might liquidate his antagonism to the people whose dirt he handles.

And then he proposes a way of sorting occupations. We might conceive of a classification of occupations involving dirty work into those in which it is knit into some satisfying and prestige giving definition of role, and those in which it is not.

Read that once more as what it is. It is not a description of an injustice. It is a specification.

Hughes is not saying the physician deserves more and the janitor deserves less. He is saying the physician's trade has built a definition of its own role large enough to contain the worst part of the work, and the janitor's trade has not built one. The dirt is the same. The container is not.

A missing container is a gap in the equipment. And gaps in equipment can be closed. This section is what we did about it.

Be precise about the deficiency first, because it is not what people assume.

It is not that the work is unskilled. It is not. Render substrates differ, and using the wrong chemistry on the wrong one will take the finish off a wall. Block paving has a structural logic that can be destroyed by a pressure washer in the hands of somebody who does not understand joint sand. Working at height has a regulatory framework that exists because people used to die regularly. There is a great deal to know.

It is not that the work is unimportant. Water in a wall is one of the most expensive things that can happen to a house.

It is not even that the work is poorly paid, which varies.

What is missing is a definition of the role. A doctor has one. A solicitor has one. An electrician, and this is the useful comparison because it is a trade, has one, and it was built. A competence framework, a certification scheme, a notification regime, a body that maintains standards and can remove people from them. Nobody was given that. It was assembled over decades, largely because electricity kills people and somebody decided that mattered.

Exterior cleaning has none of it. There is no qualification. There is no register. There is no agreed body of knowledge. Anybody may buy a pressure washer on a Saturday and be trading on a Monday, and a customer has no way whatsoever of distinguishing that person from somebody who has done it for fourteen years.

And that absence is the reason a star rating had anywhere to land. Section Nine described a number stepping into the place where a community's judgement used to be. This is the other half of the explanation. It also stepped into a place where a professional definition had never been built at all.

Porter told us that mechanical objectivity comes into its own where there is no expert consensus to satisfy outsiders. There is no expert consensus in this trade. There is not even a body that could form one.

So we built the smallest possible version of it. We are not a professional body and we are not proposing one. We are a two person sole trader in a village in Northamptonshire, and what we could build was small. We built it anyway.

A written method. What we use on which substrate, at what dilution, at what dwell time, and what we do not use. Why a soft wash and not a pressure washer on render. Why a particular approach to block paving joints. This is written down, not carried in somebody's head, and it can be shown to a customer who asks.

A case log. Individual jobs, documented. What the property was like, what was diagnosed, what was done, what it cost, what the result was, and, where relevant, what did not work. Photographed. Published.

Corrections, published. When we get something wrong, including in these six articles, we record it and leave it visible. This article has a whole section for it and that section is longer than we would like.

A stated position on evidence. No figure without a parent. No claim we cannot source. No number from a supplier who benefits from the number.

And these six articles, which between them attempt to say what is actually happening to small trades in this country, with the evidence attached.

What that is, and what it is not.

It is not a rebrand. We clean windows, gutters, render and driveways. We are a cleaning business and this is a cleaning business's website. Nothing here is an attempt to be something else, and anybody who reads it as one has misread it.

It is not a claim to be better than anybody. There are excellent operators in this trade who have never written anything down and never will, and their work is not worse than ours. The case log is not evidence of superior cleaning. It is evidence of a recorded practice, which is a different claim and a smaller one.

And it is not a marketing device, though we are aware it functions as one, and that saying it is not is exactly what somebody would say if it were. The honest position is this. We would have written it anyway, it has cost more hours than any marketing activity could justify, and the reader is entitled to discount all of it. What cannot be discounted is that it is checkable. Every source in these articles is named. Every figure has a parent or is flagged as not having one. Go and look.

What it is, is Hughes's container, built at the only scale available to us. A definition of the role substantial enough that the dirty part of the work sits inside something coherent instead of standing on its own.

Since this is the section where it belongs, here is the work itself.

Window cleaning. Residential and commercial, on a recurring cycle, pole fed and filtered, with ladder work where the property requires it.

Gutter clearance and inspection. Clearing, and telling the customer what the gutter is actually doing, which is often the more valuable half. A gutter is the cheapest building fabric protection a house has and the one most reliably ignored.

Render and rendered wall cleaning. Soft wash methods chosen for the substrate, on the understanding that render can be damaged by the wrong approach and frequently is.

Block paving and driveway restoration. Cleaning, and the re establishment of what was there.

Soffits, fascias and the roofline. The trim, and the junction between wall and roof.

Residential and commercial, across Northampton and West Northamptonshire, from The Firtrees, six Wood Lane, Hartwell. Two people. One van. A round.

Each of the six articles describes a system that decides something about a small business without knowing anything about it.

Article One. A security and sustainability apparatus built for organisations of a different size. Article Two. Financial infrastructure that assesses a business through a proxy. Article Three. A reviewing culture in which a single dissatisfied person outweighs two hundred satisfied ones. Article Four. A chargeback system in which the merchant is the defendant. Article Five. Automated decision making arriving in places nobody agreed to. Article Six. A marketplace measured by people who sell into it.

The common structure is a decision made about you by something that has no relationship with you, and that therefore substitutes a number for a judgement.

There is no small business answer to that. We are not going to pretend there is, and Section Fifteen will not pretend either.

But there is something, and Hughes identified it in 1958 without knowing he was doing it. If a trade has no definition of its role large enough to hold the difficult parts of the work, then any passing system can supply one, and the one that gets supplied will be a number. The alternative is not to defeat the number. It is to have something of your own that the number is visibly thinner than.

A written method. A case log. Published corrections. Sources you can check.

None of that changes a rating. All of it means the rating is not the only account of what happened.

THIRTEEN. WHERE WE WERE WRONG

Every one of these six articles ends with a section like this one. This is the longest of them, and there is a reason for that which is not flattering and is worth stating plainly.

This version of Article Six was built by going back and reading twenty six sources in full rather than in extracts. That process found a great many errors. Almost all of them were ours, almost all of them were in our own research files rather than in the published version, and almost all of them were the same three or four mistakes repeating.

We are publishing them grouped by kind rather than listed one by one, because the pattern is more useful than the inventory.

One. We wrote from memory instead of from the page. This is the big one. It accounts for more of what follows than everything else combined.

We identified a book from a failed extraction. Our copy of Robert Putnam's Bowling Alone has a corrupted font map, so a plain text extraction returns apparent gibberish. On that basis we recorded it as Putnam's 1995 journal article. It is the 2000 book, and the title page says so. Decoding the fault takes about ten seconds.

We located an idea in the wrong book. We recorded Simmel's blasé attitude as belonging to his 1903 essay, The Metropolis and Mental Life. It is developed in The Philosophy of Money, published in 1900, three years earlier, and developed there as a consequence of money, which is not the emphasis of the essay at all. We had written from memory of the famous version rather than from the book. The version we needed was the one we had not read.

We attributed a quotation to the wrong book by the same author. Giddens's line about locality being substantially dissolved in circumstances of modernity. We had it without a work or a year. It is The Consequences of Modernity, 1990, not Modernity and Self Identity, 1991, which is the source of his definition of ontological security. Two books, quoted one page apart in the thesis we were reading, easily merged.

We nearly built a section on a book we do not own. Section Seven was planned around Erving Goffman's front region and back region. Those come from The Presentation of Self in Everyday Life, published in 1959. We hold Stigma, from 1963, and we do not hold the other one. The concept is famous enough that we could have produced a confident paragraph about it without noticing we had never read the source. The section was rebuilt without him.

And we dismissed a book from its most famous word. We had recorded Cory Doctorow's book as a coinage rather than an argument, and advised against using it. Read in full, the machinery in it, network effects, switching costs, the collective action problem, turned out to be the best available explanation of why a small firm cannot leave a platform, and far more useful than the word.

Five errors, one cause. In every case we had a confident recollection of a well known text standing in for the text. A famous idea is the most dangerous kind to cite, because it feels like knowledge.

Two. We got figures attached to the wrong people.

Section Three's entire argument is that a figure must be traceable to whoever produced it. We failed that test twice, on that section's own two headline figures.

The ninety per cent. We wrote that Stokes and Wilson report a survey of more than three hundred owner managers. They report it. They did not conduct it. The parent is Stokes and others, 1997.

The two thirds. We credited the finding that nearly two thirds of high growth firms rate recommendations most effective to the same two authors. It belongs to Blackburn and Stokes, 2003. A different study, a different year, a different sample.

Both figures sit in one paragraph of one paper, and we let the paper's authors absorb credit for other people's research.

This is the error the article exists to warn about, committed by the article, on the article's own evidence. A figure that passes through one honest report and comes out attached to the reporter is exactly how a number loses its parent. We did not have to be careless for it to happen. We only had to not check.

And a phrase went the same way. Our working notes recorded the friendly face of enterprise as Goss's phrase. It is Jack and Anderson's own. They write, we also detect. And the word is capitalism, not enterprise. Wrong owner and wrong word, in a note written to correct a different error.

Three. We wrote down page numbers that were not page numbers.

Four times, we recorded the page of the PDF file as though it were the page of the book. Porter's poor rhetorician passage was logged at PDF page nineteen. It is page four. Hochschild's definition of emotional labour was logged at PDF page twenty. It is page seven. Her passage on class and gender was logged at PDF page thirty three. It is page twenty.

Three of those were correctable by looking at the running head at the top of the page, which is where a book keeps its page number.

The fourth is not correctable at all. Doctorow's book, in the copy we hold, is a reflowed electronic edition with no page numbers anywhere in two hundred and ninety two pages. We had recorded page fourteen and page fifteen. Those are positions in a file. There is no way to convert them, so he is now cited by chapter, which is honest, rather than by a page, which would not be.

The rule, stated once and applied everywhere. A PDF page is a note to ourselves. A book page is a citation.

Four. We got words wrong inside quotation marks. Small individually. Not small as a class, because a quotation mark is a promise.

Hughes. We had, a door which can be firmly shut. He wrote, firmly closed.

Porter. We had, the appearance of being fair and impartial. He wrote, fair and impersonal. That one is not cosmetic. Impartial means without bias. Impersonal means there is no person there, and Porter's entire argument is about the disappearance of the identifiable decision maker, not the removal of their prejudices. Our version softened his claim into an ordinary one.

And one that came from outside. Part of this article's research was audited by a second artificial intelligence system, which returned a quotation from Hill and colleagues reading, as unemployment went up, people allocated smaller portions. Those words are not in the paper. The substance is. The negative correlations between unemployment and ordinary household spending are real and are reported. But the sentence was a compression presented inside quotation marks. It is now a paraphrase, as it should always have been.

Five. We nearly printed an explanation the source does not contain. We want to describe this properly, because the near miss is more instructive than any of the completed errors.

The same audit offered an account of difficult customer behaviour built on a concept called fluid compensation. That somebody feeling powerless in one area asserts control in another, such as over a tradesman.

It is plausible. It sounds like psychology. It would have been a good paragraph. It is not what the source says.

Mandel and colleagues define fluid compensation as addressing a self discrepancy by affirming the self in a domain distinct from the domain of the self discrepancy. It is self affirmation theory. The worked examples are writing about a personally valued characteristic, and choosing an attractive product. It describes what a person does for themselves in an unrelated area. There is no interpersonal dominance limb to it in that paper at all.

We caught it by going and reading the definition.

Consider what would have happened if we had not. This article, whose central charge is that plausible sounding terms and figures get substituted for judgement, would have handed readers a scientific sounding explanation of rude customers that the cited research does not support. The failure would have been the article's own subject matter, performed by the article.

The behaviour is real. There is a published term for it, Christopher Lovelock's jaycustomer, from 1994, which describes it and does not pretend to explain it. Section Eight uses that instead, and the smaller claim is the true one.

Six. A second opinion is not a second check.

Part of this research was audited by a second artificial intelligence system, and it earned its place. Four findings in this article came from it that we did not have, and two of them are load bearing.

It also produced page numbered quotations from Mark Granovetter's The Strength of Weak Ties, stating, in the audit itself, that it knew the paper well enough not to need it pasted in. Our copy of that paper contains one thousand two hundred and ninety three characters, all of them on the JSTOR cover sheet. Pages two to twenty two are images and return nothing. Those quotations came from recollection, not from a document.

It produced five statistics from Bowling Alone. Our copy has no surviving digits.

And it produced a sentence attributed to Richard Sennett that is not in the file it was said to be in, with a publication year that disagrees with the one our source gives.

None of those entered the article. But the failure is the identical failure to our own in part one above. A confident recollection of a famous text standing in for the text. Which tells us something worth recording. It is not a machine problem or a human problem. It is what happens to anybody who is sure enough not to look.

Seven. We invented a division of labour that does not exist.

This one is the most recent, it is the worst in kind, and it was caught by the only thing that ever catches anything. Asking.

Two sections of this article, Eight and Ten, were drafted with a sentence in them saying that a second person handles the evening calls and messages. It was written as a settled fact about how this business runs. It appeared in both sections. It was on its way into a Word document.

It is not true. Every call and every message to this business has always been answered by one person, and there has never been a rota.

Worse. A further passage had been drafted, applying Hochschild's finding that emotion management has historically been one of the offerings women trade for economic support, to the division of work in this household. It was going to be offered for approval.

There was nothing to approve. We had constructed a household arrangement that does not exist and then built a sociological reading on top of it.

Note what this is, exactly. It is the same failure as the one in part five above. A plausible theory arriving before the facts and being allowed to supply them. Except that this time the invention was about a real person, in a document that would have been published under a real name.

It was caught because we asked, and because the answer came back with a correction attached. Both sentences are removed. The passage is withdrawn entirely, not held for a decision, because there is no decision to make.

Eight. And our own figure was simply wrong. Recorded in Section One, and repeated here because it belongs in this list. We published that internet sales were twenty seven point four per cent of all retail sales in Great Britain in July 2026. The Office for National Statistics figure is twenty eight point three per cent. No excuse, no mechanism, no interesting lesson. We wrote down the wrong number.

Nine. And we stretched a finding past what it could carry. Also recorded in Section Five, and also belonging here.

The first version of this article used the lipstick effect to argue that a forty pound window clean is the small indulgence a household keeps when money is tight. The study behind that, Hill and colleagues, 2012, is about women's spending on beauty products, driven by mate attraction. Five experiments, and the authors are careful about the scope throughout. Applying it to a household's decision about its guttering was using an evolutionary psychology finding as a general purpose explanation for anything that survives a downturn.

The part that does transfer is narrower and is now what we use. Economic pressure does not reduce discretionary spending uniformly. It redirects it.

And the same audit that produced the other findings proposed reintroducing the original error in a better sentence. That the forty pound window clean is the lipstick for the house. It is a good line. It is the exact thing this correction exists to prevent, and it was declined.

Four things we are taking from all this.

A famous source is the most dangerous kind. Every identification error above involved a text we were confident about. We were never once wrong about an obscure source, because we always looked those up.

A theory that fits will supply its own facts if you let it. Twice, the fluid compensation gloss and the household arrangement, a plausible reading arrived before the evidence and started filling in what it needed. Neither was caught by being careful. Both were caught by going and asking.

An extraction that returns nonsense is a fact about the extraction. Not about the document.

A figure that has passed through one honest report will arrive attached to the reporter. Every time, unless somebody checks.

And the only thing that catches any of it is going back to the page. Not care, not expertise, not a second opinion. Going and looking.

We would rather publish a section this long than one that reads well.

FOURTEEN. WHAT WE REFUSED TO PRINT

A document that shows its working should show what it left out, because the things left out are a better guide to how it was made than the things kept in.

These are the refusals. Some are about evidence we could not stand behind. Some are about material that was perfectly good and belonged somewhere else. Two are about things we could have printed and chose not to.

A study we could not read. Mark Granovetter's The Strength of Weak Ties, from 1973. It is one of the most cited papers in social science and it is directly relevant to an article about how work arrives through networks of acquaintance. We had the citation, and we had a PDF of it.

The PDF contains one page of readable text, and that page is the JSTOR cover sheet. Pages two to twenty two are images. Not one word of Granovetter's argument is in the file. We checked twice.

We could have written a paragraph about weak ties from general knowledge, attributed it to him, and nobody would ever have known. We do not cite what we have not read.

And it cost us nothing, which is worth saying because it is the interesting part. Section Three is carried instead by Stokes and Wilson, whose subject is British small firms rather than job seekers in Massachusetts in 1973. The evidence is closer to the subject and the section is better for the substitution. The famous citation was decorative.

A famous book's numbers. Robert Putnam's Bowling Alone, from 2000. Our copy has a corrupted font map. The prose can be recovered by correcting the fault. The digits cannot. Decoded sentences read, between the early s and the early s.

Putnam's book is celebrated for its statistics. We have used his argument and not one of his figures, because we do not have his figures, and a number retyped from memory of a famous book is exactly the kind of thing Section One exists to object to.

There is an accident in this that landed the right way up. The use we needed from Putnam was his caution. That nominal membership is not involvement, that a count may be measuring the wrong thing. A source we cannot take numbers from, used to make a point about the limits of numbers.

A sentence we could not find. The research audit described in Section Thirteen produced a quotation attributed to Richard Sennett, cited to a 1999 publication, page seventy five. We searched the source it was said to be in. It is not there. And our source cites Sennett's relevant book as 1998, not 1999. The sentence may well be genuine. We cannot confirm it, so it is not in the article. The place it would have filled is filled instead by a passage we can verify, from the same thesis, and which is better.

A word. Cory Doctorow's book is in our source list under its real title, which contains a word we do not print on this site. A citation must be accurate, so the title appears in the sources exactly as published. It appears nowhere else.

That is not squeamishness dressed as policy, and here is the test of it. We use the book heavily. Network effects, switching costs, the collective action problem, and the observation that users cannot help but take one another hostage. All of it is in Section Nine, quoted and attributed by name.

What we do not use is the coinage, and the reason is not the word. It is that the mechanism explains why a particular tradesman cannot do anything about his situation, and the coinage only names the outcome. We took the part that does the work. The catchphrase is the part of that book least in need of our help.

Every platform name in Section Nine. No lead generation platform is named anywhere in this article. Article Four set the rule. Describe the mechanism, and name a company only where there is documented conduct to attach to the name. We have no documented conduct by any lead platform. What we have is a structure, lean asset ownership, non portable review histories, contractors classified out of employment, and the structure is what Section Nine is about.

Naming a company we merely have feelings about would be the behaviour these six articles were written to object to.

TikTok is named in Section One, and that is not an inconsistency. TikTok has published two incompatible figures about its own business, in material that surfaces together. That is documented and anybody can check it. The rule was applied identically in both cases and produced different results, which is what a rule is supposed to do.

A passage about what people would rather not be seen buying. Hughes has a passage on how people seek certain services anonymously, using the size of a city to obtain things that are legitimate but embarrassing, from practitioners they would not want to be seen with by their own social circle. It is true, it is interesting, and it opens directly onto material that belongs to Article One, which deals with privacy and with what a business is entitled to know about the people it serves. Putting it here would have started an argument this article has no room to finish. Left where it belongs.

Marx, on piece work. Nick Srnicek, describing platforms that pay contractors by the task, quotes Marx to the effect that piece wages are the form of wages most in harmony with the capitalist mode of production. It is directly on point. It is verified, on page seventy seven of the book we hold. We are not printing it.

Not because it is wrong, and not because it is Marx. Because a sentence like that arriving in Section Nine would change what kind of document this is. A reader would stop weighing an argument about lead platforms and start deciding whether they agree with a political position. Section Nine's argument works without it and would be weakened by it, and we would rather be read than agreed with.

The same ruling covers a related thread in Jack and Anderson's paper, which argues that the enterprise culture served to re establish the standing of capitalism after the war. One attributed sentence in Section One, and no more.

A famous name that was not earning its place. Shoshana Zuboff's The Age of Surveillance Capitalism was planned into Section Nine and was cut while that section was being written. The plan allowed her one naming and one definitional quotation. Written out, she did not earn them. Srnicek's own footnote already tells the reader that the connection between platform economics and surveillance exists. The section was long. And every function Zuboff was there to perform was already being performed by somebody the section was using.

Adding a famous name to a section that does not need one is the behaviour Section One condemns. So she came out.

A very good line about lipstick. Section Five records that the first version of this article over extended a study about women's spending on beauty products into a general claim about households. During this rebuild, the research audit proposed reintroducing it in a better sentence. That the forty pound window clean is the lipstick for the house.

It is a genuinely good line. It is also precisely the error the correction exists to prevent, arriving in improved packaging. Declined.

We are recording it because a rejected error that comes back better dressed is the hardest kind to refuse, and because we would have printed it if Section Five had not already been written.

A figure that measures the wrong thing. A widely circulated statistic reports UK lipstick demand up one thousand and six per cent. It comes from a price comparison site and it measures search interest. Search interest is not sales. Printing it as evidence about spending would be wrong, and it has been printed as evidence about spending in a great many places. Not used. It is the single clearest example in this research of a number that means something other than what it is used to mean.

And everything we have ever seen on a job. This is the largest category and it has no examples, which is the point.

Nothing from fourteen years of standing outside other people's houses appears in this document. Not named. Not anonymised. Not composited from several. Not with the street changed. Not as a general impression of the sort of thing that sometimes happens.

Section Eight needed an illustration of what somebody working around a home comes to know, and of the discretion that goes with it. That illustration comes from Everett Hughes's 1958 interviews with Chicago janitors. People who agreed to be studied, in a published book, seventy years ago, in another country.

We used a dead sociologist's research so that we would never have to use a living customer's house. That is the most deliberate decision in the article and it is the one we would least want misunderstood.

Eleven refusals. Three because we could not verify the source. Three because the material belonged to another article or another argument. Two because a famous name was not earning its place. One because a number measured something other than what it was being used for. One because a word does not go on this site. And one because it is not ours to tell.

Not one of them was refused because it was inconvenient to our argument.

That was the test we applied, and a reader is entitled to check it. Every item above either weakened the evidence, belonged elsewhere, or was somebody else's to disclose. If we had suppressed something that undermined us, this list would be where it should have appeared, and we would like to be asked about it.

FIFTEEN. WHAT WE THINK

Everything up to here has been evidence, or our observations marked as observations, with the join between them left visible. This section is the opinion. It is clearly labelled, it is ours, and a reader who has followed the evidence is perfectly entitled to reach a different one.

The claim. The marketplace being described to a one van business is not the marketplace it operates in. Not a distorted version of it. Not an exaggerated version. A different market, with a different logic, described using the same words.

A local trade runs on proximity, repetition and personal trust. Work arrives because somebody who lives nearby told somebody else who lives nearby. The unit of growth is a street. The asset is a round, and a round is several hundred separate arrangements held in several hundred separate households, none of them written down anywhere that matters.

Platform reach runs on the opposite of all three. It is geographically indiscriminate. It does not know or care that we cannot service an address ninety miles away. It is non recurring by design, because a completed transaction generates no further rent. And it is rented by the click, which means access is a cost that resets rather than an asset that accumulates.

These are not two versions of one market. They are two economies with incompatible logics, and the data describing one is being sold as a map of the other.

That is the whole claim. Everything in Sections One to Fourteen is the working.

What follows from it, and what does not.

It does not follow that online is a waste of money for everybody. Plenty of businesses sell things that travel, to customers who never need to be met. For them the map is the territory and the advice is sound.

It does not follow that the platforms are dishonest. Section Nine went to some length to show that every mechanism described there is ordinary competition behaving exactly as economics predicts. No villain is required and none was found.

It does not follow that referral is morally superior. It is a method. It happens to be the method that nine in ten small firms report using, and to be the one the apparatus describes least.

What does follow is narrower and, we think, more useful. A small local trade being advised to reorganise itself around figures it cannot check, in a market whose size cannot be established within a factor of thirty, is being asked to take on faith something that the people asking have not established either.

Six articles, six subjects, one shape. A security framework built for organisations of another size. A financial system that assesses a business through a proxy. A reviewing culture in which one dissatisfied person outweighs two hundred satisfied ones. A dispute process in which the merchant is the defendant. Automated decisions arriving where nobody agreed to them. And a marketplace measured by the people who sell into it.

In every case, something decides about you, has no relationship with you, and therefore substitutes a number for a judgement, and then presents the number as though nobody had decided anything.

And Section Eleven establishes the part that matters most, which is that this is not something happening to small businesses. It happens in three directions at once. To us, through a format with no field for rain, for a hospital appointment, or for the health of the person doing the work. To the customer, who once judged a frontage themselves and now has two words and an average, a capability lost without anybody noticing it going. And to the trade, which narrowed not because the work got simpler but because the vocabulary for describing it did, and the equipment followed the vocabulary down.

Three parties, one structure, and none of the three can see the other two.

Theodore Porter named that in 1995 and we have leaned on him throughout. His finding is the one we would most want a reader to take away, so here it is once more in his words rather than ours. Quantification is a way of making decisions without seeming to decide.

Not because numbers are bad. Because reaching for one is evidence about the state of trust in a situation. Where there is a community that knows you, its judgement is used. Where there is not, a measurement stands in, and the measurement is not an upgrade on the neighbour's word. It is what gets built when there is no neighbour.

We opened with Everett Hughes, and we want to close with him, because the argument has come round.

Hughes's claim in 1958 was not that low prestige work is dignified, though he plainly thought it was. It was methodological. That prestige builds a front which hides the inside of things, that in trades of less prestige the core may be more easy of access, and that studying them is a matter of finding the best possible laboratory.

If that is right, and we think fourteen years of evidence says it is, then what has been assembled here is not a cleaning company's opinion about marketing. It is a report from the place where the mechanism is visible.

Everything described in this article happens to larger businesses too. It happens behind procurement departments, account managers, agency relationships, and the general reluctance of successful organisations to describe how they really operate. Down here there is none of that. There is a man on a ladder, a van with a name on it, a rating administered by strangers, and a bill for a lead that may or may not convert.

No front. Nothing between the mechanism and the person it is operating on. That is what makes this a poor position to be in and a good position to write from.

What are we actually doing about it? Not much, and we would rather say so than end on something rousing that is not true.

We cannot change a rating system. We cannot make a review history portable. Section Nine explains why nobody will. We cannot cause the marketing profession to describe word of mouth accurately, and we cannot establish the size of a market that four organisations with an interest in the answer cannot agree on within a factor of thirty.

What we can do is the thing Hughes's 1958 finding actually specifies, and it is small. A trade with no definition of its own role large enough to hold the difficult parts of the work will have one supplied by whatever system is passing, and the one that gets supplied will be a number.

So. A written method. A case log. Corrections published rather than overwritten. Every source named, every figure with a parent or flagged as having none, and these six articles.

None of that beats the number. All of it means the number is not the only account of what happened, and that a customer who wants to know more than a decimal between four and five has somewhere to go.

That is the whole of it. It is not a strategy and it is not a solution. It is a container, built at the only scale available to two people and a van, because the alternative was to let something else define what this work is.

And one last thing about the evidence. Every claim in these six articles can be checked. The sources are named, dated and page referenced. Where we could not verify something, we said so. Where we got something wrong, Section Thirteen records it, and that section is longer than any of us would have liked. Where we refused to print something, Section Fourteen says what and why.

We would rather be checked than believed. If any of it is wrong, we would like to know, and it will go in the corrections.

Matthew Kenneth McDaid. Shining Windows. The Firtrees, six Wood Lane, Hartwell, Northampton, N N seven, two H G.

SOURCES

Everything cited in this article, with edition, page and licence position. Where we read a work at second hand, that is, where somebody else reported it and we did not obtain the original, it is listed separately and marked, because the difference matters.

Quotations from copyright works are used under the quotation exception, section thirty, Copyright, Designs and Patents Act 1988.

A note on how these sources have been used. Most of what these writers argue appears in this article in our own words with their names attached, not in quotation marks. That is deliberate.

An idea is not owned. Hughes's argument that the low prestige trade is the better laboratory, Porter's argument that numbers replace judgement out of weakness, Simmel's argument that the manner of acquisition enters the thing acquired. None of those is anybody's property, and stating them in our own words, with credit, is the ordinary business of reading.

The wording is owned. So where we have used somebody's exact sentence it is short, it is in quotation marks, it is attributed in the body text with a page number, and it is there because the phrasing is itself the argument. Where a claim is given in our words, the claim is still theirs.

Works read in full and cited directly.

Everett C Hughes, Men and Their Work, published in 1958 by The Free Press of Glencoe, Illinois. Used, chapter three, Work and the Self, pages forty two to fifty five, and chapter eight, Dilemmas and Contradictions of Status, pages one hundred and two to one hundred and fifteen. Chapter eight was first published in the American Journal of Sociology, volume fifty, March 1945. Cited at pages forty eight, forty nine, fifty, fifty one, fifty two, one hundred and three, one hundred and eleven, and one hundred and twelve. In copyright.

Erving Goffman, Stigma, Notes on the Management of Spoiled Identity, published in 1963 by Prentice Hall, Englewood Cliffs, New Jersey. Used, chapter one. Cited at pages two, three, four including footnote three, fourteen and fifteen. In copyright.

Theodore M Porter, Trust in Numbers, The Pursuit of Objectivity in Science and Public Life, published in 1995 by Princeton University Press. Cited at pages four and eight. In copyright.

Georg Simmel, The Philosophy of Money, third enlarged edition, edited by David Frisby, translated by Tom Bottomore and David Frisby from a first draft by Kaethe Mengelberg. Published by Routledge. First published 1978, second edition 1990, third edition 2004. Cited at pages two hundred and fifty seven and two hundred and fifty eight. There are two copyrights here and they are different. Simmel's German original of 1900 is public domain. The Bottomore and Frisby English translation is copyright Routledge and is a protected literary work in its own right. We have therefore given Simmel's argument at length in our own words and quoted the translation only twice in the whole article, where the English phrasing is itself the point.

Nick Srnicek, Platform Capitalism, published in 2017 by Polity Press, Cambridge. Cited at pages forty three, forty nine and fifty, seventy six and seventy seven, and one hundred and thirteen. In copyright. The passage quoted from pages seventy six and seventy seven contains a printing error in the original, which we have reproduced as printed.

Cory Doctorow. The title of his book appears in our full written source list and nowhere else in this article, for the reason given in Section Fourteen. Cited by chapter, not by page. Our copy is a reflowed electronic edition carrying no page numbers in two hundred and ninety two pages. A position in a file is not a page in a book. In copyright.

Michael H Goldhaber, The attention economy and the Net, published in First Monday, volume two, number four, on the seventh of April 1997. Cited by section heading, as the article carries no pagination. The New Natural Economy. Illusory Attention. Money and Attention. Wealth and Property Take New Forms, Too. And, A Point Worth Repeating, Though Not Too Often. Open access, but copyright is retained by the author. Open access governs reading, not reproduction, and he is quoted short and attributed like everybody else. Retrieved and verified in full on the sixteenth of September 2026.  https://firstmonday.org/ojs/index.php/fm/article/view/519

Arlie Russell Hochschild, The Managed Heart, Commercialization of Human Feeling, published in 1983. Cited at page seven, page seven footnote, page eight and page twenty. In copyright.

Donald Horton and R Richard Wohl, Mass Communication and Para Social Interaction, Observations on Intimacy at a Distance, published in Psychiatry, volume nineteen, in 1956. Cited at article level, without page numbers, because our copy is an unpaginated reprint. The page range of the original is still to be confirmed. In copyright.

John Suler, The Online Disinhibition Effect, published in CyberPsychology and Behaviour, volume seven, number three, in 2004, by Mary Ann Liebert Incorporated. Cited at page three hundred and twenty one. In copyright.

Brooke Erin Duffy, Not Getting Paid to Do What You Love, published by Yale University Press, New Haven. We hold a thirty two page publisher preview, not the book. Nothing beyond those pages is attributed to her anywhere in this article. Cited at preview pages eleven and thirteen. In copyright.

Irwin Altman and Setha M Low, editors, Place Attachment, published in 1992 by Plenum Press, New York and London, as volume twelve of Human Behavior and Environment, Advances in Theory and Research. Used, chapter one, Place Attachment, A Conceptual Inquiry. Cited at page seven. In copyright.

Naomi Mandel, Derek D Rucker, Jonathan Levav and Adam D Galinsky, The Compensatory Consumer Behavior Model. We hold an author manuscript. Cited to manuscript pages and section numbers throughout, and the body text says so. Journal, year, volume and published pagination still to be confirmed. In copyright.

Sarah E Hill, Christopher D Rodeheffer, Vladas Griskevicius, Kristina Durante and Andrew Edward White, Boosting Beauty in an Economic Decline, Mating, Spending, and the Lipstick Effect, published in 2012 in the Journal of Personality and Social Psychology. Abstract quoted. Volume, issue and page range still to be confirmed. In copyright.

David Stokes and Nicholas C Wilson, Entrepreneurship and marketing education, time for the road less travelled, published in the International Journal of Entrepreneurship and Innovation Management. Both authors, Kingston University. We hold an author manuscript whose header carries a placeholder in place of the volume and issue. Internal manuscript pagination is intact and is what we cite, pages eight, nine and ten. Published volume, issue, year and page range still to be confirmed. In copyright.

Sarah L Jack and Alistair R Anderson, Entrepreneurship education within the enterprise culture, Producing reflective practitioners, published in 1999 in the International Journal of Entrepreneurial Behaviour and Research, volume five, number three. Cited at pages one hundred and fourteen, one hundred and fifteen and one hundred and sixteen. In copyright.

Matthew Adams, doctoral thesis. Used as the source for Giddens, Thompson, Bauman, O'Brien and Lasch, all of whom are quoted through it and are listed as second hand below. Cited at pages sixteen, thirty seven and three hundred and twenty six. Full title, institution and year still to be confirmed. In copyright.

Robert D Putnam, Bowling Alone, The Collapse and Revival of American Community, published in 2000 by Simon and Schuster, New York. Argument only. No figure from this work appears in this article. Our copy carries a corrupted font map which destroys every numeral in the file. Section Fourteen explains. In copyright.

Official statistics. The Office for National Statistics, series J four M C, internet sales as a percentage of total retail sales. The series is published at o n s dot gov dot uk, under business industry and trade, retail industry, time series, j four m c, d r s i. The bulletin is Retail sales, Great Britain, July 2026, published at o n s dot gov dot uk, under business industry and trade, retail industry, bulletins, retail sales, july 2026. Dataset released on the twenty first of August 2026. Open Government Licence version three point zero, free to reproduce with attribution.  https://www.ons.gov.uk/businessindustryandtrade/retailindustry/timeseries/j4mc/drsi  https://www.ons.gov.uk/businessindustryandtrade/retailindustry/bulletins/retailsales/july2026  http://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/

Works cited at second hand. These are named in the article because their findings are used. We have not obtained the originals. Each is reported by a source above, and the body text says so wherever it appears.

Reported by Stokes and Wilson. Stokes and others, 1997, the survey of three hundred plus owner managers. Blackburn and Stokes, 2003, the high growth firms study. East and others, 2000, recency and referral. Stokes and others, 2002, what owners believe about which customers refer. And Kotler, 2003, and Brassington and Pettitt, 2006, whose page counts are as reported by Stokes and Wilson and were not independently re counted by us.

Reported by Jack and Anderson. Goss, D, 1991, Small Business and Society, published by Routledge, London. The initial is confirmed as D. The forename is not confirmed and is not printed. Anderson, 1995, the omnibus word formulation. And Curran and Burrows, 1987, page one hundred and eighty, on the remoralisation and recapitalisation of capitalism.

Reported by Mandel, Rucker, Levav and Galinsky. Higgins, E T, 1987, Self discrepancy, A theory relating self and affect, in Psychological Review, volume ninety four, pages three hundred and nineteen to three hundred and forty. Wicklund, R A, and Gollwitzer, P M, 1981, Symbolic self completion, attempted influence, and self deprecation, in Basic and Applied Social Psychology, volume two, number two, pages eighty nine to one hundred and fourteen. Wicklund and Gollwitzer, 1982, Symbolic Self Completion, published by Lawrence Erlbaum Associates, Hillsdale, New Jersey. Cutright, K M, 2012, The beauty of boundaries, When and why we seek structure in consumption, in the Journal of Consumer Research, volume thirty eight, number five, pages seven hundred and seventy five to seven hundred and ninety. That is not to be confused with Cutright and others, 2011, When your world must be defended, in the same journal, volume thirty eight, number one, pages sixty two to seventy seven, which is a different paper on a different subject. Rucker, D D, and Galinsky, A D, 2009, Conspicuous consumption versus utilitarian ideals, How different levels of power shape consumer behavior, in the Journal of Experimental Social Psychology, volume forty five, number three, pages five hundred and forty nine to five hundred and fifty five. And Rucker and Galinsky, 2008, whose full reference is still to be confirmed.  https://doi.org/10.1016/j.jesp.2009.01.005

Reported by Harris and Reynolds. Christopher Lovelock, 1994, the term jaycustomer, whose full reference is still to be confirmed. The source is Lloyd C Harris and Kate L Reynolds, The Consequences of Dysfunctional Customer Behavior, Cardiff University.

Reported by Matthew Adams. Anthony Giddens, 1990, The Consequences of Modernity, page one hundred and three, on locality and ontological security. Anthony Giddens, 1991, Modernity and Self Identity, pages sixty one and two hundred and forty three, on ontological security defined. Those are two different books, and Section Thirteen records our having merged them. E P Thompson, 1993, Customs in Common, published by Penguin, London, cited at page three hundred and fifty seven. Zygmunt Bauman, 1998, Globalization, The Human Consequences, page eighty six. M O'Brien, 1999, page thirty eight. And Christopher Lasch, 1984, The Minimal Self, page thirty one.

Reported by Brooke Erin Duffy. Blogging dot com, a survey of one thousand bloggers. A commercial operator in the market it surveyed. Self selected sample. Not peer reviewed. This is stated in the body text alongside the figure.

Consulted and not used. Mark Granovetter, 1973, The Strength of Weak Ties. Not cited. Our copy contains one page of readable text, which is the JSTOR cover sheet. Erving Goffman, 1959, The Presentation of Self in Everyday Life, published by Doubleday, Anchor Books, New York. Not held and not cited. The reference is confirmed via Goffman's own footnote in Stigma, page forty four, note two. Section Thirteen records that Section Seven was originally planned around it. Richard Sennett. Not cited. A sentence attributed to him could not be located and the publication year was in dispute. Shoshana Zuboff, The Age of Surveillance Capitalism, 2019. Planned into Section Nine and removed during writing. Section Fourteen explains why.

Sources belonging to the other articles, read for this research, used elsewhere, and not duplicated here. Hu, Pavlou and Zhang, on self selection bias in online product reviews, which belongs to Article Three. Baumeister and others, Bad Is Stronger Than Good, which belongs to Article Three. And Brown and others, which belongs to Article Five.

Every figure and every passage in these six articles belongs to exactly one of them.