Why asking customers what they want can give you worse answers

A customer says a professional programme felt expensive and the business writes down: price is too high. Those are not the same claim. The first is evidence about what the customer said; the second is an explanation of why they did not buy, and it may or may not be right.

That gap is where customer research often goes wrong. Interviews are useful because customers can describe what happened around a decision, what they were trying to achieve, what worried them, what they compared and what got in the way. Problems start when a comment about the experience is treated as a diagnosis, or when a hypothetical answer is treated as evidence of what somebody will actually do.

When I am trying to understand a buying decision, I want the customer's account of what happened, but I also want enough context to judge what that account establishes.

Ask about what actually happened

If I am trying to predict whether somebody will spend money or change behaviour, I usually put more weight on what they have already done than on what they say they might do.

Asking "Would this help?" invites somebody to imagine a future and usually makes it easy to say yes. Asking what happened the last time they tried to solve the same problem gives you something more concrete: what triggered the search, what they tried, what they compared, what they paid for, whether they postponed the decision and what eventually happened.

Past behaviour still does not tell you exactly what someone will do next. Circumstances change, and genuinely new products or first-time purchases may not have a neat precedent. It is still a useful reference point because the person had to deal with real constraints rather than imagine them in an interview.

This is also why general enthusiasm does not tell me very much on its own. Someone can say a problem matters and still have lived with it for years without trying to change it. Another person may sound less enthusiastic but have already spent time or money trying several alternatives.

The right research also depends on the decision you are trying to make. Understanding why somebody bought is different from testing whether they can use a page or asking how well a programme worked afterwards. I would rather ask a narrower set of questions for a specific decision than collect a large pile of customer feedback and hope the answer is hiding somewhere inside it.

Customers can tell you a lot about the situation

There is a dismissive version of this argument which says customers do not know what they want, so you should not listen to them. I do not find that useful.

Customers can tell you a great deal about the situation they were in: what changed, what they were trying to do, which alternatives they considered, what felt risky, where they got confused and the language they naturally use to describe the problem. That is often exactly the information the business is missing.

I am more cautious when the conversation jumps from that evidence to asking the customer to prescribe the solution.

If someone says they struggled to work out whether a training programme would fit around their job, that tells us something important about the buying decision. It does not tell us whether the programme should become shorter, the curriculum should change, another format should be added or the sales page simply needs to explain the existing requirements earlier.

Those choices still need judgement about the product, delivery, economics, positioning and what the business is trying to achieve. Customer research can improve that judgement without replacing it.

The same applies when a customer asks for a different format or way of participating. The request may point to a real constraint, but taking it seriously means understanding that constraint before deciding whether the offer itself should change.

Keep the answer separate from the diagnosis

When customer language points towards a problem, I still want to compare it with what people actually did. Depending on the decision, that might mean transactions, support records, email behaviour, participation or what happened after purchase.

I sometimes look at the same question another way. We might offer two different videos from an email, or frame sessions around different topics and see which people actually click or watch. That gives me a different signal from asking in a survey or poll which topic they say they want.

A click isn't perfect evidence either. Email clicks can include bots, open rates are noisy, session timing can affect attendance, and a stronger title can attract more clicks without proving the underlying topic matters more. The point is not to replace customer research with behavioural data. It is to see whether different imperfect signals point in the same direction.

That matters because customer explanations can be completely sincere and still incomplete. Someone may remember one concern most clearly, while the wider evidence shows that another part of the decision was already weakening earlier. A support question can reveal uncertainty without proving what caused it. A request for a different format can reveal a constraint without proving that creating the format would be commercially sensible.

For a buying journey, I therefore treat interviews and customer comments as part of the evidence rather than the whole answer. Depending on the question, I may also want transactions, support records, email behaviour, onboarding information or what people actually did after they encountered the offer.

I try to keep the customer's words and behaviour intact before turning them into an explanation or recommendation. That makes it easier to see what the evidence actually supports and where my own interpretation begins.

In practice, I keep those things separate long enough to see whether the customer's words, their behaviour and the rest of the evidence are pointing in the same direction. When they agree, confidence rises. When they disagree, the disagreement is often more useful than either signal on its own.