Your price sorts who shows up before it does anything to your revenue. Set it wrong and you have not merely left margin on the table — you have recruited the customers who churn fastest, file the most tickets, and never expand, while repelling the serious buyers you wanted. Price is a filter first and a number second, and the cheapest customers usually arrive with the worst version of the problem you solve.
Most founders treat price as an output. You build a cost model, add a margin, benchmark the competitor, and read a number off the bottom of a spreadsheet. That process answers "what can we charge and stay solvent" while ignoring the question that decides whether the company works: who does this number let in the door. The spreadsheet is downstream. Selection happens first.
Willingness to pay is a signal, not a nuisance
Start with the thing pricing models treat as noise: the spread in what different buyers will pay for the same product. Standard practice sees that spread as a problem to be captured — segment, fence, extract more from the high end. Fine. But the spread is also information, and it is information about the buyer, not just the wallet.
Willingness to pay tracks problem severity. The person who will pay a lot to solve something is, on average, the person for whom the problem is expensive, urgent, and central. The person who will only pay if it is nearly free has, on average, the shallow version of the same problem — a mild annoyance, a nice-to-have, a thing they abandon the moment the discount ends or a marginally cheaper option appears.
Now chain that forward, because this is where the selection logic bites. Problem severity tracks retention: the deeper the problem, the more it hurts to leave. It tracks expansion: deep problems grow, spawn adjacent needs, pull in more seats and more usage. And it tracks inversely with support cost per dollar — not because severe-problem customers ask for less, but because they are trying to get real work done and their requests compound value, whereas the shallow-problem customer treats a cheap tool as disposable and expects it to be frictionless anyway.
Put the three together and you get an uncomfortable result. Your cheapest customers are disproportionately the ones with the worst unit economics: lowest revenue, highest support cost, fastest churn, zero expansion. You did not stumble into that cohort. Your price recruited them. A low number is a casting call, and the people who answer it are, by construction, the ones for whom the problem was never worth much.
The number is a claim about seriousness
The mechanism runs the other way too, and this is the half founders underweight because it is invisible in the funnel. A low price does not only attract the wrong buyer. It repels the right one.
A serious buyer reads price as a quality signal, because in most markets it is a reliable one. When someone evaluating a solution to an expensive problem sees a suspiciously low price, the rational inference is not "great deal." It is "this is not built for a problem like mine." Low price reads as low seriousness, thin support, a hobby project that will be abandoned, a vendor who does not understand what is at stake. The buyer with the severe problem cannot afford to bet their outcome on something priced like it does not matter. So they self-select out, silently, and you never see them in your pipeline. You see the cohort your number invited.
This is not a psychology quirk to work around. It is the same logic as why your best marketing often looks like waste: a price you could not sustain if the product were bad is itself a costly signal. It certifies seriousness precisely because a fly-by-night operator could not hold it. Underprice, and you strip out the signal — you look exactly like the low-quality option you are not, and the buyers most able to tell the difference are the ones who walk.
Trust-scarce markets turn the dial up
I build for emerging markets where institutional trust is weak — cash-on-delivery commerce, where the buyer will not prepay because the enforcement scaffolding a Western founder assumes simply is not there. In that setting the price-as-signal effect is not a second-order refinement. It is a primary force.
With no reliable court, no trusted ratings infrastructure, no brand you already know, the buyer has fewer instruments for inferring quality — so each remaining instrument carries more weight, and price is one of the loudest. A price that reads as cheap does not say "affordable." It says "not serious, probably gone in six months, do not hand this one your trust." I have watched underpricing actively destroy credibility with the exact merchants a product most needs: the serious operators reason that a real tool for a real business would not be priced like a toy, and they route around you to something that looks like it has skin in the game. In a low-trust market, a low price is not a discount. It is a confession.
The general principle: the weaker the other quality signals available to your buyer, the more selection work your price does whether you intend it or not. B2B enterprise, regulated industries, high-stakes personal decisions, frontier markets — all of them lean harder on price as a proxy, which means all of them punish underpricing more severely.
Discounting is the dynamic version of the same error
If price selects the customer, a discount re-runs the selection at a worse setting. Every point you knock off does two things at once: it lowers the revenue on that deal, and — the part that gets ignored — it changes who accepts. A discount preferentially closes the buyer who was sitting at the margin of paying at all, which is to say the price-sensitive, shallow-problem, low-loyalty buyer. You are not winning a customer you would otherwise lose. You are over-weighting your book toward the cohort most likely to leave and most likely to demand.
I have argued the mechanics at length in Discounting Doesn't Buy Customers, so I will hold to the selection point: a chronic discounting habit does not merely compress margin, it slowly rebuilds your customer base out of the people your full price was correctly screening out. Then you look at your retention numbers, conclude the product has a stickiness problem, and go build features for customers who were never going to stay. The discount laundered a selection decision into what looks like a product failure.
Read your churn as a pricing diagnosis
That last move is the practical payoff, so make it explicit. When a company has a persistent churn problem or a support load that scales faster than revenue, the reflex is to treat it as a product problem — ship more, fix onboarding, add a success team. Sometimes that is right. But a chronic churn-or-support problem is also a signature of a pricing problem, and the two are hard to tell apart from the inside because both feel like "the customers are unhappy."
Here is the tell. Segment your churn and your support tickets by the price the customer actually paid — realized price, discounts included. If your worst-retaining, highest-support, never-expanding customers cluster at the bottom of your price distribution, you do not have a product problem in that cohort. You have a selection problem, and no feature you ship will fix it, because you are serving people who were correctly filtered out and then let back in through the cheap door. The fix is not downstream in the product. It is upstream in the number.
Most teams never run this cut. Blaming the product feels more actionable, because the product is a thing you can go change on Monday, whereas admitting the price recruited the wrong base means questioning a decision baked into every contract you have signed. An unexamined price is a standing instruction to your funnel about who to bring you, and if it is set wrong it keeps bringing you the same losing cohort no matter how good the product gets.
Decide the customer, then derive the price
Invert the order of operations. Do not price the product. Decide the customer, and let the price fall out of that decision.
Name the buyer you want — the severity of problem they have, the retention and expansion they are capable of, the trust environment they live in, the support intensity they will actually require. That specification implies a price band: the number that selects that buyer and screens out the ones who would wreck your unit economics. Set the price to hit the buyer, not to clear the spreadsheet. Then — and only then — build the business that price can fund. A price that selects serious buyers can pay for the high-touch support serious buyers expect; a price set for self-serve volume cannot, and running one on the economics of the other is how companies quietly die while blaming the product.
So the pricing question was never "what should we charge." It was "who do we want to walk through the door, and what number puts them there." The finance falls out afterward. Get that order backwards and the spreadsheet will balance perfectly, right up until you notice it balanced around the customers you never wanted.
Your price is talking to the market whether you meant it to or not. The only choice is whether it is calling the right people.