Nobody briefed it to come back with a no.

The meeting went well.

That is the part worth sitting with, because none of what follows is about work done badly. It went well. Somebody walked the room through the thing, the screens did what they were supposed to do, nothing broke in front of an investor, and two people said it looked great. One of them said it twice.

You left pleased. Everybody left pleased.

Then, somewhere in the fortnight after, you noticed that you could not actually do anything with what had happened in that room. Not because the feedback was poor. Because there was no feedback, in the sense of information that could have changed your plan. There was approval, which is a different substance entirely, and approval does not break down into decisions.

Six weeks and a piece of your runway, and you know precisely what you knew before.

Approval is the only answer in the room.

Here is the mechanism, and it has nothing to do with the quality of the build.

Think about what you actually asked of those people. You asked them to look at a thing and understand it. They did. They understood it, they could see what it was for, and the honest response to having understood something is to say so. That is what this looks great means. It means I followed it.

There was no other answer available. Not because the room was being polite, though it was. Because nothing in the exercise created a position anybody could take against it. You cannot refuse a thing you have merely been shown. There is nothing there to refuse.

So the artefact did not fail to produce a decision. It was never carrying one. It arrived correct, it arrived on time, it matched the brief in every particular, and it was incapable of the one outcome you needed.

Two briefs, and the line between them.

Briefed to be understood. Make the idea legible. Show how it works, show what it would feel like, make the thing in my head visible to people who are not in my head. That is a reasonable thing to want, and it is what almost every proof of concept in an outsourced build is asked for. It ends in approval, because approval is its ceiling.

Briefed to extract a commitment. Put this in front of eleven people who would have to pay for it, and come back with what they did. Not what they said. What they did, when asked to part with something they would rather keep.

The second brief has two possible endings, and that is the whole point. It can come back with a yes. It can also come back with a no.

The no is the more useful of the two. A yes tells you to carry on, which is what you were going to do anyway. A refusal tells you something available no other way, at a price you can still afford, in a month where changing course costs you a fortnight rather than a year. Nine refusals out of eleven is not a failed experiment. It is the most valuable thing anybody has handed you all year, and it arrives dressed as bad news.

A pricing tier and a prototype brief have that much in common. Both decide, in advance, what range of answers is available. One to a customer, one to a room.

Diagram · the answers each brief makes possible

The left column is what a room can give a thing it has merely been shown. The right column is what it can give a thing that asked it for something.

Nobody is cutting a corner.

Now the part that is nobody’s fault, and it is worth being precise about, because the reflex is to be annoyed at the people who built it.

A brief is a document written to be delivered against. That is its function. A supplier is judged, correctly, on whether what arrived matches what was described, and that judgement is the thing protecting you from the wrong ones. It is the basis of the whole arrangement.

So look at what happens to a firm that takes the second brief seriously. They build it, they put it in front of eleven buyers, and nine of them decline. Then they come back into a room with you and report that.

What does that meeting look like? It looks like the engagement failed. It gets written up as a disappointing outcome. The invoice sits in your accounts next to a result nobody wants to discuss, and the next founder who calls that firm for a reference hears about it.

Every incentive points at a result that can be signed off. The room responded well is a deliverable. Nine of eleven declined is a problem, and it is a problem the firm will be judged to have caused.

I built and exited an agency, so let me put that where it belongs rather than somewhere convenient. We would have built what the brief asked for and built it well, and nobody on our side of the table was ever going to propose an exercise whose best possible outcome made our own work look like a failure. Not out of self-protection. Because that brief has to be written by the person who benefits from the no, and that person is you.

What you actually bought.

So price it honestly. You did not buy a proof of concept. You bought a meeting that could not have gone badly.

The cost of that is not the invoice. It is the meeting you could have had instead: the one that was allowed to go badly, in month four, while the plan was still cheap to change. You bought the version where nothing is risked and therefore nothing is learned, and you paid the same money for it.

And the loss compounds in a way an invoice never shows. The approval you collected becomes evidence. It goes into the next deck, it gets repeated to the next investor, and it hardens into a thing everybody around you now believes, on the strength of a room that was never in a position to disagree. You are not back where you started. You are further down a road with more weight behind you and the same amount of evidence, which is none.

A proof of concept that cannot come back with a no and a build quote that cannot describe year two are the same kind of document. Both complete, both correct, and neither ever asked to carry the answer you actually need.

An instrument that cannot return a negative.

Here is the way of thinking about it that makes the fix obvious.

A proof of concept is an instrument. It exists to take a reading. An instrument that cannot return a negative is not measuring anything, it is confirming. A thermometer that only ever reads warm is not a thermometer. A test that cannot fail is not a test, it is a ceremony with equipment in it.

That is the whole standard, and it is a low bar that almost nothing clears. Before you commission the thing, ask what reading it is capable of returning. If every outcome available to it is some shade of encouraging, you have specified a ceremony.

The ones that work are built around the refusal. Not show them the product, but ask them for something they would rather not give, and tell me what happened. Money, a signature, a date in a diary, an introduction to the person who actually signs. Something with a cost attached, so that a yes carries weight and a no is available at all.

We have built one that way, briefed to come back with signed intent rather than warm words, real enough that a buyer could see what it would earn them, and judged on what came back rather than on how the room reacted. It is a different object from the first day, and the difference sits in the brief rather than in the build. N°50 covers what the cheapest version of that looks like when there is almost no budget to spend on it.

The sentence, before anybody builds.

One exercise, and it costs nothing.

Write down the sentence you want somebody to say at the end of the meeting this thing exists for. In their words, not yours. In quotation marks, as a sentence a real person would actually utter out loud.

Then look at what you have briefed, and ask whether that sentence is even possible to say inside it.

This looks great is possible inside almost anything.

Yes, put me down for the first twenty is possible inside almost nothing that gets built, because nothing in the artefact asks for it and nothing in it can receive the answer.

If the sentence you wrote cannot be said, you have not briefed a proof of concept. You have briefed a presentation, and it will come back and tell you, at length and very convincingly, exactly what you already believed.