- Name a price and charge someone in the next thirty days. The thing you are testing is the value, not the polish, so the product does not need to be finished.
- A free pilot answers one question: will somebody accept something for nothing. A paid one tells you whether the problem has a budget, who signs, how long the buying takes and what you were compared to.
- Model the twenty per cent move upwards as well as down. A twenty per cent cut needs twenty five per cent more customers just to stand still. A twenty per cent rise can lose one customer in six and still be ahead.
- Nobody on the supplier's side of the table is paid to ask whether the number is correct. That question is due before the billing gets built, not after.
Every pre-seed room has the same sentence in it, and it always arrives in the reasonable tone of voice. We will figure out pricing later. After the product is right. After we know what it is worth. After the round. Nobody in the room argues, because it sounds like discipline. It is one of the most expensive sentences a founder says out loud, and the problem with it is not the logic. It is the calendar.
Open your build plan and find the line that produces the first pound of revenue from someone who is not related to you. Read the date next to it. For most founders I meet that date sits after the round they are raising, which means the single strongest piece of evidence they could put in front of an investor is scheduled to arrive after the moment they needed it. Later is not a decision to price carefully. Later is a decision to raise without the evidence a price would have produced.
Price is the cheapest experiment you own.
Every other experiment on your roadmap costs a sprint. Test the onboarding, rebuild the flow, ship it, wait for numbers. A price costs an email and about four days. It is the only experiment in the business with a build cost of zero, and it is the one nobody schedules.
It is also the only one that tests the right thing. Usage tests interest. Sign-ups test curiosity. Retention tests habit, eventually, if you wait long enough to see it. A price tests value, because value is the number at which somebody would rather part with money than lose the thing. Every conversation you have after you name a number is a test of four things at once: whether the value is real, whether this person is the buyer, whether this segment has budget, and whether what you built is a product or a favour.
What a free pilot cannot tell you.
A free user answers exactly one question, and it is not a useful one. Will they accept something for nothing. Almost everybody will, especially if they like you, and pre-seed design partners usually do like you. Six enthusiastic free partners is a slide. It is not evidence, and the round can tell the difference.
A paid pilot answers four questions in the same four days:
- Is the problem worth a budget line? Not interesting, not clever. Worth money that was allocated to something else last quarter.
- Who signs? The distance between your user and that person is the shape of your whole go to market, and you cannot see it until money is involved.
- How long does the buying take? Procurement, security review, legal. That elapsed time is your sales cycle, and your sales cycle is priced in runway.
- What did they compare you to? The comparison tells you which category the buyer filed you under, which is your positioning, arriving as a fact rather than a workshop.
None of those four have an answer while the thing is free. And the most useful reply you get will not be yes. It will be somebody asking whether there is an annual discount, because people do not haggle over a price they think is absurd. They go quiet. The founder who never replies was the strongest signal in your last four board updates, and the silence is the finding.
There is a cleaner way to see it. N°61 argued that every deliverable on a statement of work should be marked O or E: an output you will own, or evidence a real user generated. A paying pilot is an E, the most fundable one available to a pre-seed company. A free trial is almost always an O in disguise, a thing you built, being used, producing nothing an investor can weigh.
The arithmetic nobody runs upwards.
The three tier structure is the least interesting part of pricing and it is already well understood. The cheap tier exists to kill the price objection and is not really meant to be bought. The expensive one exists to make the middle look reasonable. The middle tier is the product: it is what your onboarding assumes, what your support model rests on, and what every default in the build quietly serves. So there is only one number worth arguing about, and the argument is arithmetic.
Here is a worked example with the assumptions on the table, so you can redo it with your own. Twenty customers on a £400 middle tier. That is £8,000 of monthly revenue and £96,000 of ARR. It costs you £60 a month to serve each customer and about £900 to acquire one. Now move the middle number twenty per cent in each direction and watch what happens to the things a round actually asks about.
The symmetry is the part worth keeping. Cut twenty per cent and you need twenty five per cent more customers just to arrive back where you started. Raise twenty per cent and you can lose one customer in six and still be ahead on revenue. Every founder runs that number downwards constantly: a discount to close, a founder rate, a friend price. Upwards, almost never. If you have never run it up, you did not price the product, you flinched at it.
And revenue-neutral is not profit-neutral. The twenty five per cent more customers you won with the discount each cost money to serve, each need onboarding, each open support tickets, so the same £8,000 arrives with less of it left over and a longer wait to earn back what you spent acquiring them. The cut looks free on the pricing page and shows up as margin and payback everywhere else. That is the same conversation as N°47, arriving a year earlier than most founders have it.
Then there is what the number says before it earns anything, which founders consistently underrate. Too low reads as unfinished. It invites the buyer who chose you on price, who will leave on price, and it anchors every renewal conversation you will ever have with them. Too high without proof does not offend anybody, it simply stalls: procurement asks for a reference customer at that price and you have none. The number is a positioning statement before it is a revenue number, and it is very hard to un-say.
The number nobody on the other side of the table is paid to check.
Tiers are not a pricing page. They are metering, limits, permissions, billing logic, upgrade and downgrade paths, proration, failed payment emails and what happens the day somebody stops paying. Every one of those is a build decision, and they are all made the moment the price is. A build partner will build exactly the pricing the plan describes, properly, and will be right to. Nobody on that side of the table is paid to ask whether the number is correct, whether the tiers match how the buyer actually buys, or whether the company is about to spend a quarter building billing for a price no human being has ever said yes to.
I can say that without it being a complaint, because I built and exited an agency at Atomise. We built billing to specification and we built it well. The specification was accurate. Accurate is not the same as accountable, and the question of who had agreed to the number going into it belongs to somebody sitting on the founder's side of the table. In most pre-seed engagements that chair is empty.
Earlier this week I made the same argument from the other end of the invoice: a blended day rate hides the shape of the team you are buying, and nobody on the founder's side ever checked the shape. This is that failure on the revenue side. A number set by looking sideways at the nearest competitor, with nobody on the founder's side asking whether anybody would pay it. Same gap, two sides of the same invoice, and the same missing person.
The rules, plainly.
- Name a price before you build the billing. The billing is a quarter of engineering built on top of an assumption. Test the assumption first, it costs an email.
- Charge somebody in the next thirty days, even if the product is not finished. You are testing the value, not the polish. Nobody has ever refused to pay because a settings page was ugly.
- Never run a free pilot without a written end date and a price it converts at. A pilot with no conversion date is not an experiment, it is a donation with reporting attached.
- Model the twenty per cent move up before you model it down. If a discount is the only direction you have ever tested, you have never actually tested your price.
None of this asks you to know the right number. You will not know it, and the point of an experiment is that you do not have to. It asks you to stop treating the price as a decision waiting on the product, when the product is the thing that has been waiting on the price. Price is usually the load bearing belief nobody has tested, which makes it exactly the kind of assumption N°57 is about, and it is by some distance the cheapest one to check.
So pick the three most engaged people on your free plan. Send them a number and a date this week. Then let a stranger's answer replace your opinion, which is the whole of what an experiment is, and the only version of this you can put in a deck.