How to spot a pricing validation mistake before it crosses the line

A pricing mistake is rarely a wrong number. It is a right number read as proof of something it cannot prove. So each entry here has three parts: what was done, what it was taken to mean, and what was actually established.

The last section is different in kind. It is about tests that produce good data by treating a customer badly, and those are not mistakes of reasoning.

What to take away

Mistakes of method

Asked people what they would pay.

Quoted a range instead of a number. Taken to mean: flexibility, and a wider chance of a yes. Actually established: that the bottom of the range is your price. The reader hears the low number and every subsequent conversation starts there.

Softened the number in the pause. Taken to mean: good salesmanship. Actually established: nothing. The silence after a price is the instrument, and filling it deletes the reading. If you discount before they have spoken, you have tested your own nerve.

Tested price and offer in the same week. Taken to mean: faster learning. Actually established: an unattributable result. Change one thing at a time, as the offer testing pillar sets out, or accept that the number tells you nothing about which change caused it.

Quoted higher to the ones who looked wealthy. Taken to mean: sensible targeting. Actually established: a correlation between your judgment of a person and the price they heard, which cannot be untangled afterwards. Assign price points in rotation or you have no comparison.

Mistakes of reading

Everyone said yes.

Nobody said yes. Taken to mean: the price is too high. Actually established: one of four things, and the difference decides the next move: they cannot justify the amount, they do not want it much, they are not the person who decides, or they were never in your segment. Separating those is what the pricing questions is for.

Two out of five bought at the top price. Taken to mean: the market bears the top price. Actually established: very little. Move one event and the conclusion flips, which is the check to run on any small sample before acting. The reasoning is in the discovery counts.

A competitor charges a certain amount. Taken to mean: the market price. Actually established: what appears on a page. Their real prices include discounts you cannot see, contracts you cannot read, and costs you do not have. Public artifacts are worth studying and they are not a price, which is the argument in competitor validation.

The price was accepted, so the business works. Taken to mean: the arithmetic is fine. Actually established: one number that somebody agreed to. Whether it works depends on the cost of delivering and the cost of acquiring, and the split between one-time and recurring costs decides the second, as the startup cost guidance sets out.

Mistakes of structure

Chose monthly billing because everyone does.

Bundled to hide a price. Taken to mean: a better offer. Actually established: that nobody can now tell you which part they wanted. A bundle is a fine thing to sell and a poor instrument to learn from.

Set a price at the cost floor. Taken to mean: competitive. Actually established: survival, and nothing else. The floor includes delivery, support, rework, losses, and the cost of acquiring one customer through a channel you can repeat. A price at it leaves nothing to be wrong with.

Introduced a discount to close a slow month. Taken to mean: a temporary push. Actually established: a new price. Everyone who paid full price learns it, and the discount becomes the number the next conversation starts from.

The line a test may not cross

These are not errors of inference. They are choices, and they are available, and they should not be taken.

Fake scarcity. A countdown that resets, a limited number of something not limited, a closing date that does not close. It works once, on the people who did not notice, and it tells everyone who did notice exactly what your other claims are worth.

A pre-sale you cannot deliver. Taking money for something you have not confirmed you can ship is not a price test, it is a loan you did not disclose. If you do pre-sell, the delivery window you state is an obligation, and the order and shipment rule requires notice and a prompt refund when you cannot meet it.

Claims that make the price look justified. Time saved, money returned, results achieved. A claim needs a reasonable basis before it runs, and price pressure is exactly when unsupported ones get written.

Charging different people differently and hiding it. A price ladder run openly is a legitimate test: everyone quoted a number is offered the thing at that number, and gets it at that number. The version where the higher price is presented as the only price, or where the lower one is withdrawn afterwards, is a different activity.

A subscription that is hard to leave. Auto-renewal buried, cancellation routed through a call, terms shown after payment. This is the subject of the rule on negative option offers, and it also destroys your own evidence, because a person who cannot cancel never tells you why they wanted to.

The check that catches most of them

After any pricing exercise, write one sentence: this test establishes that, at this price, in this structure, these particular people did or did not commit. Then read it back and delete any word in your conclusion that the sentence does not support.

Most of the entries above disappear under that test, because most of them are a conclusion one clause wider than the evidence. The rest of the method is in pricing validation.

Common questions

Is asking about price ever useful?

Asking what somebody currently pays for something adjacent is useful, because it is about the past. Asking what they would pay is not, because it is about a future in which nothing is at stake.

How many refusals do I need?

Enough that the refusals sort into groups. A handful of refusals that all say different things means you are talking to different segments, which is itself the finding.

Can I test a price without a product?

Yes, if the commitment is real: a deposit, a signed order, a scheduled start. The commitment has to cost the person something, and you have to be willing and able to honor it.

What if my price is already public and wrong?

Change it, and honor the old number for anyone who already saw it. Testing an increase on new customers only is clean and slow, and it is much better than the version where existing customers discover the change from a bank statement.