Covers the gap with language
Just launched, and honestly I'm blown away by the response.
The feedback so far has been incredible.
This is everything I've learned in three years, in one place.
DM me if you want in.
Free article — Behavioral economics / B-04
What counts as proof when nobody has bought from you yet?
Someone has replied to your post with four words: “Any proof this works?” There is no hostility in it. They are asking the obvious question, and you notice that answering honestly means answering with nothing. Nobody has bought it. Nobody has reviewed it. Your follower count is three digits and you do not know how many of those are real.
Most people take one of two exits here. They ignore the reply, or they round upward. “Early feedback has been great.” Nothing happens that day, so next week it gets rounded a little further. This slope has no floor, and the people who slide down it are almost never dishonest by temperament. They are just out of material.
There is a third exit. You can build honest proof from a standing start, without a single buyer and without borrowing anybody else's credibility. The short version of this article is that you are not short of evidence. You are sitting on evidence you have not yet put into a shape that counts.
About fifteen minutes to read, with a five-minute inventory at the end. Free, in full. There is nothing to fill in on this page and nothing is collected. If you finish it holding two or three pieces of proof you could publish today, it has done its job.
Before the theory, one judgement
Two sellers, both with nothing to show. Which one do you believe?
Same number of sales, which is zero. Same number of reviews, which is zero. Comparable audiences, comparable products. The only difference is what sits in the pinned post.
Leaves the gap where it is
Nobody has bought this yet. I have no customer feedback to show you.
Instead: pages 1 to 27 are public, in full, at the link. Read them before deciding anything.
I also ran the method on my own business and published the result, including the two fields I could not fill in.
What it does not do: it will not get you an audience, and it does not come with support.
B has just written three things against their own interest — no sales, an unfinished self-assessment, and a list of what is missing. Most readers still find B more credible. That is worth sitting with for a second, because the usual explanation (honesty is charming) is not the interesting one.
Here is the interesting one. Everything B published is something an empty product cannot produce. You can only make page 27 public if page 27 exists. You can only publish the result of running your own method on yourself if you have a method and you ran it. You can only name the two fields that defeated you if you actually got stuck in them.
Proof is anything a seller with nothing would lose by copying.
That definition is narrower than the everyday sense of the word, and it is the reason it is useful. Take the thing you are about to publish, imagine the emptiest competitor in your niche doing exactly the same thing, and ask whether it costs them anything. If it does, you have evidence. If it costs them nothing, you have decoration, and decoration is what readers scroll past.
Why claiming does not work
“I'm good at this” is not information. It is noise with a nice font.
In 1973 Michael Spence published a paper about the job market whose structure is almost exactly your situation. Applicants know how capable they are. Employers do not. An applicant saying “I am excellent” conveys nothing, because a weak applicant says the identical sentence at the identical cost.
Spence's answer was that the difference travels through actions whose cost differs between the two types. Not through better claims. Through something the weaker party would find too expensive to imitate. What matters is the asymmetry in cost, not the intrinsic usefulness of the action.
Biology has a parallel argument. In 1975 Amotz Zahavi proposed that a display can be reliable precisely because it is costly to carry — surviving while burdened is itself the message. It has been contested and refined for fifty years and it is not a law you can lift into commerce. As a way of thinking about why your loudest competitor is also your least believable one, it is hard to beat.
Anything anyone could say costs nothing to say, and is therefore worth nothing to hear.
The case for giving away the part you could sell
Phillip Nelson made a related argument about advertising in 1974. The content of an advertisement may be unverifiable, but the fact that a firm spent heavily on it can still be informative: heavy spending only pays back if buyers come back, and buyers only come back if the product held up. The expenditure is the message.
You are one person selling from a social feed and you have no advertising budget. The structure still transfers, because you do have something expensive: the work itself. Publishing a real, usable section of it is catastrophic if the rest is thin and recoverable if the rest is good. That asymmetry is the whole mechanism.
This is why “the first 27 pages are public” lands differently from “the content is excellent.” It is not generosity that the reader is responding to. It is that you have visibly accepted a cost that only makes sense if you expect the remainder to hold up. Some readers will take those pages and leave. That risk is not a side effect of the signal; it is the signal.
The obvious caveat: this only works if the part you published is genuinely useful on its own. Publishing a thin excerpt does not signal confidence. It demonstrates thinness, at your own expense, to exactly the people you were trying to reach.
The ladder
Five rungs you can climb before anyone has paid you anything.
Sales figures and customer praise sit near the top of this ladder and you cannot reach them yet. The five rungs below are all available today. They run from cheapest and easiest to imitate, to most expensive and hardest.
State the specification
SpecificationPage count. File formats. How long it takes. What is included. What is not. Facts you can confirm by opening your own files.
It is skipped constantly because it feels beneath the pitch. While it is missing, a reader has nothing at all they can check, and no amount of adjectives above it changes that.
Show the actual thing
DemonstrationThe table of contents. Screenshots of interior pages. A page after it has been filled in. A recording of the work happening.
This replaces “it is good” with the reader's own judgement. A reader who has judged for themselves no longer needs anyone else's word for it, which is convenient, because you do not have anyone else's word.
Hand over a piece of it
SampleOne complete opening section. One worksheet. Step one of eleven. Choose by completeness, not by length.
Cut where something finishes, never where something is interrupted. A sample that stops mid-thought has not been given to anyone. It has been dangled, and readers can tell the difference instantly.
Publish the boundary
ConstraintWhat you do not do. What is not included. Who this is wrong for. Situations where it will not help.
This is the rung people skip hardest, and it is the least imitable one on the ladder. An empty product cannot afford a boundary, because once the boundary is drawn there is nothing left inside it.
Run it on yourself, in public
Self-applicationTake the method you are selling, apply it to your own business, and publish the result — including the parts that did not resolve.
It is the strongest rung for a dull reason: you cannot apply a method you do not have. And a self-assessment where everything worked reads as fiction, which means the honest version is the only one that functions.
You do not need all five at once, and the order is load-bearing. Rung 5 without rung 1 reads as an eccentric personal blog post. Rung 1 under rung 2 under rung 3 is a structure, where each rung is corroborated by the one beneath it.
A track record is what arrives after this ladder, not what you need in order to build it.
Rung 4 deserves one extra note. Publishing what you do not do reads like pure downside, and it has a second effect that arrives after the sale: it fixes the buyer's expectations before they pay, and expectations you set are expectations you can meet. It is your least copyable pre-sale proof and your most effective post-sale complaint reduction, in one sentence.
A worked rewrite
One pinned post, rebuilt on the ladder.
An example, invented for this article rather than taken from anyone. The seller offers one-to-one consultations on clothing.
Papering over the gap
Personal styling consults, now open ✂️
So grateful for how this is taking off already!
“I should have booked this months ago” — the response has been amazing.
Slots are going fast, DM to grab one.
Rungs one to four, in public
Styling consults. I have run zero paid sessions so far and I have no client feedback to show you.
What it is: one 45-minute video call, and a shopping list of up to nine items we build together on the call.
What it is not: I do not audit your whole wardrobe, I do not come shopping with you, and I do not buy anything on your behalf.
Who it is wrong for: if you already have a uniform that works, there is very little here for you.
So you can judge before booking: here is the same list format, made for myself, with the budget floor and the two things I got wrong (4 images).
The rewrite argues against itself three separate times — zero sessions, three exclusions, one group told not to book. It also replaced an invented quotation with four things a reader can actually check. Count what a stranger can evaluate: before, nothing; after, four items.
Why “I have sold none of these” is a tactical sentence
The resistance to writing that line is real, and the usual framing of it — be brave, be vulnerable — misses what it is doing. The purpose is not to display honesty. It is to close an open question in the reader's head before that question starts colouring everything underneath it.
Leave it unanswered and the reader carries “but has anyone actually bought this” through every subsequent line. Answer it in the first sentence and the question is settled, badly but definitively, and their attention is free to move to the four checkable things you put directly below it.
So this is a sequencing decision, not a confession. You are supplying information in the order it is being demanded. Humility is not required and, frankly, is not what is doing the work.
When it does not work
Four conditions that decide whether a signal carries anything.
Amna Kirmani and Akshay Rao reviewed the quality-signalling literature in 2000. What is useful about a review is that it records the failures as well as the successes. Restated from the seller's side, four conditions matter.
Is the cost visible to them?
A hundred hours of work is invisible. An invisible cost signals nothing, and writing “this took me a hundred hours” does not fix it, because that sentence is free to type.
What works is a cost that is legible in the artefact — a level of detail, a number of worked examples, a completeness that could not have been assembled quickly.
Would lying actually hurt you?
Promises about things nobody can check later are cheap to make and are discounted accordingly, however sincerely you meant them.
Promises a buyer verifies within ten minutes of paying — page counts, formats, what is in the folder — are expensive to fake and are believed almost for free.
Can they even judge it?
A first-time buyer in your category has no standard to hold your sample against. Handing them the work does not help if they cannot tell good work from bad.
For that reader, give them the criteria first. Explaining what to look for is itself a signal, and a fairly costly one, since it invites them to apply it to you.
Is everyone else already doing it?
A signal everyone emits carries no difference. On a page where every listing has five stars, five stars are information-free.
Look instead for the disclosure nobody in your field makes. It is usually sitting inside whatever everyone has silently agreed to conceal.
The fourth is where the leverage is. Find the thing everyone in your niche hides — how the price was arrived at, how long it really takes, the project that went badly, the client who was a bad fit — and open one of them. Same effort as everything else on your page, and it is the only part nobody can match by tomorrow morning.
The review organises conditions. It does not supply a coefficient, and neither does this article. Nothing here tells you how much any of it moves.
Borrowed credibility
Someone else's record does not become yours by proximity.
With no record of your own, the nearest material is somebody else's. A former employer's name, a well-known person's name, a famous study's name. All three look substantial sitting on a page. They behave very differently.
“Ex-[large company]”
True, presumably. But the reader's question is what that history guarantees about this product, and a job title on its own answers nothing.
It starts working when you draw the line yourself: “I did accounts payable for seven years, so I have seen most of the ways a freelance invoice goes wrong.” Now it is information.
“Recommended by [name]”
Genuinely strong, under two conditions: they agreed to it, and if anything of value changed hands you say so in the same breath.
Both the FTC's endorsement guidance and Japan's rules on advertising that hides being advertising exist because of what happens when those conditions are skipped.
“According to [study]”
Research does not vouch for your product. This article cites six sources and none of them is offered as evidence that anything sold here is worth its price.
Citing is for showing where your thinking came from. That is a real service to the reader and it is not, by itself, proof of anything about you.
Borrowed credibility has to be paid back. Proof you built yourself does not accrue interest.
Borrowing has a second problem: it does not describe you. Thousands of people share your former employer and none of them made your product, so the more impressive the borrowed name, the further a reader has to travel from it to reach the thing they are considering. The four lower rungs are faster anyway. They need nobody's permission and nobody else can produce them.
After the first one
What you may ask a first buyer for, and what you may do with it.
Eventually someone buys. This is a hinge, and it is where a lot of sellers damage a relationship in the first hour by asking for the wrong thing at the worst possible moment.
Start from the fact: a buyer owes you nothing further. They paid for a product, not for promotional labour. “If you enjoyed it, a quick word would mean a lot!” is politely worded pressure, and the minutes after payment are when it is hardest to decline.
Ask a different question. “Where did you get stuck?” It is easier to answer, it does not require anyone to compliment you, and unlike praise it is directly usable. Praise is raw material for marketing you should not be doing yet. A stuck point is raw material for the next revision.
What you want from your first buyer is a fault report, not a compliment.
Three rules for publishing what comes back
- Ask before you publish, and specify the exact passage. One sentence of agreement. Skipping it damages the relationship with the first person who ever backed you, which is an expensive thing to spend on a quotation.
- If anything of value changed hands — a free copy, a discount, a fee, an arrangement — say so beside the quote. This is what the FTC's guidance is about, and in Japan concealing it has been treated as a violation of the representations law since 1 October 2023.
- Publish it at full length or not at all. Trimming “the first half confused me, but it clicked by the end” down to “it clicked by the end” is not editing for space. Two options removes the temptation entirely.
And when praise does eventually arrive, keep the lower rungs. One quotation is not a sample size, the reader has no idea who said it, and specification, demonstration, sample and boundary go on doing their work at a hundred quotations just as well as at zero.
This site currently holds no customer feedback of any kind, so none is displayed anywhere on it. An article arguing this position while quietly bending the rule would not be worth reading.
What this article is not claiming
Tendencies and frameworks, not laws.
Everything above is cited for direction. Stating the limits explicitly, because the citations are doing less work than citations usually pretend to.
- Spence's paper is a model of how cost asymmetry can carry information in a market with hidden quality. It is not a prediction that publishing part of your work will produce sales.
- The handicap principle is a contested hypothesis in evolutionary biology. It is used here as an analogy and is not evidence about human commerce.
- Nelson's argument concerns advertising expenditure by firms. It does not directly address an individual giving work away, and is borrowed because the structure is similar.
- The 2000 review organises conditions across many studies. No effect size from it is quoted here, and none should be inferred.
- The FTC guidance and the Japanese designation are regulatory documents. They describe what is not permitted. They say nothing about what sells.
And the limit that matters most: you can build all five rungs and still not sell anything. Proof makes a good product findable and believable. It does not make an unsuitable product suitable, and no amount of it substitutes for the product being right for the person reading.
Where the line is
Everything above can be faked, so here is what I will not do.
Praise can be invented. A demonstration can be staged. A self-assessment can be written to flatter. Since this article is effectively a manual for both versions, the refusals belong in it.
Not doing
- Writing words for a buyer who does not exist. A composite of several real people is the same act with extra steps.
- Reviewing my own work under another identity, or arranging for someone to do it as though unprompted.
- Accepting anything of value for a recommendation and leaving that out. If there is a relationship, the relationship gets printed next to the words.
- Presenting a free or discounted copy's response as an ordinary purchase.
- Listing work I was adjacent to as work I did. “Involved in” has a great deal of room in it, and that room is the problem.
- Editing a quotation until it says something the person did not say.
- Publishing a number I did not measure. Satisfaction rates, completion rates, average results.
Doing
- Saying there is no feedback while there is no feedback.
- Publishing only specification I can verify by opening the file.
- Marking anything given away as given away.
- Putting what failed in the same place as what worked, at the same size.
- Drawing the boundary at a level of detail that costs me enquiries.
- Telling a first buyer, before they answer anything, that they are under no obligation to.
The real cost of invented proof is not being caught. It is that the reason to build the real thing disappears the same afternoon.
Write “the response has been great” and the ladder stops, because rhetorically you are already at the top. Next month you write the same sentence from the same rung. The record never accumulates — not for want of time, but because a substitute was available and it was free.
Five minutes, on paper
Inventory what you can already publish.
Pick one product you think of as having no proof behind it. Five lines, on paper or in a notes app.
- 01Five facts you can confirm by opening your own files.Length, format, duration, what is included, what is not. If you cannot reach five, that is information about the product rather than about the page.
- 02What could you photograph or screenshot as it is?Contents page, interior, a finished example, the work in progress. If the reason you are not showing it is that it looks unpolished, show it. Unpolished real things beat polished descriptions more often than sellers expect.
- 03Which complete piece could you give away without regret?Not where it can be interrupted — where something finishes. The unit is whatever lets the recipient complete one thing on their own.
- 04Three things it does not do.Excluded, out of scope, not solved here. Three answers is the outline of your product, and it is the rung nobody around you is climbing.
- 05Have you run your own method on yourself?If not, that is your next piece of work. Publish it with the unresolved parts left unresolved, and rung five is built.
Most people fill three or four of the five. Which means the proof already existed and simply was not published. The reason it was not published is almost always the same: it looked too ordinary to count. It looked ordinary because you have been staring at it for months. Nobody else has seen it at all.
Nothing on this page is a form. Whatever you write stays where you wrote it — there is no mechanism here to receive it.
Where this came from
Six starting points, linked in full.
None of the thinking above is mine. Links go to the original paper, the publisher, the DOI, or the regulator's own page. Most are in English; some sit behind paywalls I cannot remove, so the full citation is given for anyone using a library.
Spence, M. (1973)
The Quarterly Journal of Economics (DOI)
The framework this article is built on: an unobservable quality can be conveyed by an action that is cheaper for the high-quality party than for the low-quality one. Borrowed for its direction, not for any effect size.
Zahavi, A. (1975)
Mate Selection — A Selection for a Handicap
Journal of Theoretical Biology (DOI)
Where the idea that a costly signal is harder to fake comes from. This is a biology hypothesis that has been argued over ever since; it is used here as an analogy and not as a law about commerce.
Nelson, P. (1974)
Journal of Political Economy (DOI)
Argues that the fact of spending on advertising can itself carry information, separately from what the advertisement says. Used here for the case where giving work away is the expenditure.
Kirmani, A. & Rao, A. R. (2000)
No Pain, No Gain: A Critical Review of the Literature on Signaling Unobservable Product Quality
Journal of Marketing (DOI)
A review of the signalling literature. Used to set out the conditions under which a signal does and does not work, rather than to claim that any of them works by a particular amount.
U.S. Federal Trade Commission
The FTC’s Endorsement Guides: What People Are Asking
Federal Trade Commission (official)
The official US guidance on endorsements and reviews: what has to be disclosed, and what an endorsement is allowed to say. Relevant to anyone publishing in English.
Consumer Affairs Agency, Japan (2023)
Consumer Affairs Agency, Government of Japan (official)
The Japanese rule that took effect on 1 October 2023. This site is operated from Japan, so this is the line it is held to. The page itself is in Japanese.
These are cited for the direction of a described tendency and for the boundaries set by two regulators. No effect size is claimed from any of them, and none of them is offered as evidence that Chapter 1 is worth what it costs. Research does not vouch for a product.
That is the end of the article
By rung four, the missing thing usually stops being proof.
Work up the ladder and most people stall at the same place. Writing what you do not do turns out to require knowing what you do — and that has to be settled before the boundary can be drawn at all.
Rung five makes it unmistakable. Running your own method on yourself is only informative if the thing being run through it has a shape. Put a vague product through a self-assessment and you get a vague self-assessment. The quality of your proof is set by the sharpness of the product's outline, not by how well you write about proof.
There is one paid thing on this site, and it is the continuation of this. Chapter 1, “Turning a positive into a product,” takes your own experience through nine fields — who it is for, what changes, what it costs, where it goes out — and ends with one design sheet in your own words. A reader PDF, a workbook PDF and the same workbook as an editable DOCX, for US$39. It is not reading. It is two to three hours of writing.
I have applied this article to that product. The Chapter 1 page carries the nine fields run against the chapter itself, with the fields I could not fill in left visibly empty — including why the price is that number rather than another one. Read it against the rest of the page before deciding. If the two disagree anywhere, that is a reason not to buy.
And to be explicit: the six papers and documents cited above do not establish that US$39 is the right price for anything. They describe how cost asymmetry can carry information. Whether the chapter is worth it is a judgement you make from its contents, and it is not one I can make from here.
Who should not buy it
Before anything else, the people this is wrong for. If any of these describe you, do not buy it.
- You already know exactly who your buyer is and what to change next when a response comes back
- You want something to read rather than something to fill in
- You are looking for a reliable way to increase your income quickly
- You are looking for audience-growth tactics or advertising operations you can use today
What the chapter does not promise
- That your product will sell
- That you will make back what you paid
- That your income will increase
Everything the chapter contains, everything it leaves out, and everything it refuses to promise is written on its own page. The only things missing from that page are sales figures and customer feedback, which do not exist yet — and the page says so.