Almost every company I have worked with or looked at has an ideal customer profile.
It exists as a slide. It has firmographics, a revenue band, a set of industries, a few buying triggers, and usually a persona or three with names and stock photography. Somebody worked hard on it. It was reviewed. It is not wrong.
The question is whether it governs anything.
A test that takes ten minutes
Ask four leaders separately, in writing, to describe the customer the company is built to serve. The head of sales, the head of marketing, the head of product, and the CFO.
Do not let them confer, and do not accept the deck as an answer.
Most of the time the four answers are recognisably related and materially different. One names a revenue band. One names an industry. One names a technical trigger such as a platform migration or a compliance deadline. One names the segment where gross margin is best.
Nobody is being difficult. Each is describing the customer their function is optimized around, and those are not the same customer.
That divergence is not a communication problem to fix with a workshop. It is already priced into decisions that have been made for years.
The operative ICP
Alongside the stated ICP there is always a second one, which I think of as the operative ICP. It is not written anywhere. You derive it.
It is what the last four to eight quarters of closed won business actually looks like, combined with whatever the pipeline is currently full of.
The operative ICP is the one the organization genuinely believes, because it is the one the organization has been rewarded for. Territory design reflects it. Compensation accelerators reflect it. The hiring profile for sellers reflects it, because the people who succeeded look a certain way and the next hires are chosen to look like them. The content library reflects it. The events calendar reflects it.
When the stated and operative profiles agree, the machine reinforces itself and growth is comparatively easy to forecast.
When they diverge, everything downstream inherits the divergence, and it does so silently.
What divergence actually costs
The costs are rarely attributed to the ICP, which is why they persist.
Marketing generates demand against the stated profile. Sales works the operative one. Conversion looks poor, and the diagnosis becomes lead quality, which is a permanent and unwinnable argument between two teams who are each doing their job correctly.
Product prioritises against the stated profile. Revenue arrives from the operative one. The roadmap slowly stops matching the customers who are actually paying, and nobody can point to the decision where that happened.
Forecasting assumes stated-profile deal shapes: a cycle length, an average value, a conversion rate. The pipeline is full of operative-profile deals with different shapes. The forecast is wrong in a way that looks random from quarter to quarter and is in fact perfectly systematic.
And the new market entry that leadership has been discussing gets planned against the stated profile too, which means it inherits assumptions the company has not actually validated anywhere.
Why it drifts
Nobody decides to let this happen. It accumulates.
Every deal teaches the organization something. A seller finds an unexpected pocket of demand and works it. It closes. Others notice. Within two quarters a meaningful share of pipeline is coming from a segment that appears nowhere in the strategy, and it is working, so nobody questions it.
That is not a failure. It is a company learning something about its market, which is exactly what you want it to do.
The failure is that the learning never gets promoted back into the stated profile. The slide stays as it was, and the gap between what the company says and what it does opens quietly, one successful exception at a time.
The uncomfortable version
Here is the part leadership teams tend not to enjoy.
When the two profiles diverge, the instinct is to correct the behaviour, on the assumption that the strategy is right and the field has drifted.
Sometimes that is true. Often it is not.
The operative ICP is built from evidence. It represents everything the market has actually told the company about where it wins, at what cost, and against whom. The stated ICP frequently represents where the company would like to be, or where it was three years ago, or the segment that made the funding case work.
Before deciding which one is wrong, it is worth taking seriously that the answer might be the document.
What to do
The point is not to force the two into agreement. It is to know which one you are running.
Derive the operative profile from closed won data. Put it beside the stated one. Name every place they differ, and for each difference decide explicitly whether you are going to move the strategy toward the evidence or move the behaviour toward the strategy.
Both are legitimate answers. What is not legitimate is leaving it undecided, because undecided means the organization keeps running two customer definitions at once, and paying for both.
An ICP is not a targeting document. It is a set of decisions about where the company will and will not spend its capacity. If it is not making those decisions, it is not an ICP. It is a slide.