Customer Signals Are Messy: Why Startup Validation Needs More Than Asking People What They Think

One of the hardest parts of assessing a startup idea is working out whether customers actually want it.

That sounds obvious. Talk to potential customers, ask them about the problem, show them your idea and see what they think.

The problem is that customer signals are rarely that straightforward.

People say things they don't mean. They complain about problems they have no intention of paying to solve. They find workarounds instead of buying products. They praise ideas because they don't want to discourage the person presenting them. And sometimes the strongest evidence of a market opportunity doesn't come from someone explicitly asking for a solution at all.

If you're trying to assess whether a startup idea has genuine demand, you need to look at a much wider range of evidence.

What are customer signals?

Customer signals are pieces of evidence that tell you something about a customer's problems, priorities or behaviour.

Some signals are very direct.

Someone might post: "Does anyone know a tool that can do X?"

That's useful. A person has identified a problem, actively looked for a solution and publicly asked for one.

But most market evidence isn't this convenient.

Someone might instead complain about spending three hours every Friday updating a spreadsheet manually. They haven't asked for a product. They may not even realise that software could solve the problem.

However, their behaviour is still telling you something.

This is why assessing customer demand requires looking at different types of signals rather than simply counting how many people say they like an idea.

Direct customer signals

Direct signals are the easiest to recognise.

These include people explicitly: asking for a product or feature searching for recommendations complaining that no suitable solution exists saying they would pay for something looking for alternatives to an existing product describing a specific unmet need

These can be strong evidence, particularly when they appear independently across different sources.

However, even direct signals need context.

Ten people saying "I'd love an app that does this" is not necessarily evidence of ten customers. The important question is what happens next.

Adjacent signals can be just as important

Some of the most useful evidence is adjacent to the product rather than directly about it.

Imagine you're considering building software that automates a particular administrative process. You might find very few people saying: "I need software that automates this process."

That could suggest there is no demand.

But then you discover discussions where people are sharing elaborate Excel templates, paying freelancers to complete the work, creating internal scripts, or complaining about how much time the process takes.

Those are adjacent signals.

Nobody is asking for your proposed product, but people are already investing time, effort or money into solving the underlying problem.

In some cases, that can be more meaningful than somebody simply saying they would like your idea.

Workarounds are evidence

One of the strongest customer signals is often a workaround.

People rarely sit around waiting for a startup founder to invent the perfect solution. They solve problems using whatever is available.

That might mean: maintaining complicated spreadsheets combining several different software tools hiring freelancers building internal systems using generic products for purposes they weren't designed for manually transferring information between systems creating templates, checklists or scripts

A workaround demonstrates something important: the problem is significant enough for somebody to do something about it. That doesn't automatically mean they will buy your product. But it provides evidence of behaviour rather than merely opinion. And behaviour matters.

What customers say and what customers do are different things

This is one of the biggest problems with traditional idea validation.

Ask somebody:

"Would you use an app that helped you save money on groceries?"

There is a good chance they will say yes. Why wouldn't they? Saving money is good. The app sounds useful. Saying yes costs them nothing. But their actual behaviour might tell a different story.

Perhaps they already have access to supermarket comparison tools and don't use them. Perhaps they repeatedly choose convenience over saving money. Perhaps they abandon budgeting apps after a week. Perhaps saving £5 isn't worth spending ten minutes entering information.

The hypothetical customer and the real customer can behave very differently.

This is closely related to the difference between stated preferences and revealed preferences.

Stated preferences are what people tell you they would do.

Revealed preferences are what their actual decisions show they value.

For startup research, both can be useful. But they shouldn't automatically be given equal weight. "That sounds interesting" is not validation

There is another problem with simply asking people whether your startup idea is good. People are generally quite nice. If you enthusiastically explain the business you've been working on for three months and then ask: "What do you think?" you have created a terrible research question.

The other person isn't just evaluating the idea. They're navigating a social interaction. They probably don't want to tell you: "I don't think anyone will pay for that." So instead you get: "That's interesting." Or: "I could definitely see people using that." Or the particularly dangerous: "I'd probably use something like that."

None of these statements require the person to actually do anything.

This doesn't mean they're deliberately lying. They may genuinely think the idea sounds good. But thinking something is a good idea and becoming a customer are completely different behaviours.

Complaints don't automatically equal demand either

The opposite mistake is assuming every complaint represents a market opportunity.

People complain about products constantly.

They complain about prices, subscriptions, interfaces, missing features, advertising and customer service. But dissatisfaction does not necessarily create switching behaviour.

Someone can spend five years complaining about the same piece of software while continuing to pay for it every month.

That tells you something too.

Perhaps switching costs are high. Perhaps competitors are equally bad. Perhaps the problem isn't important enough to justify changing. Or perhaps the existing product solves other problems well enough that customers tolerate its weaknesses.

A complaint is therefore evidence. It isn't automatically evidence that someone will buy your alternative.

Willingness to pay is another signal entirely

There is a large gap between: "I have this problem." and: "I will pay you to solve this problem."

That gap is where many startup ideas struggle.

Some problems are irritating but not economically important. Others have perfectly adequate free workarounds. Some affect people who don't control the purchasing decision. Others occur too infrequently to justify another subscription.

This is why willingness-to-pay signals deserve particular attention.

Evidence that people are already spending money — on competitors, consultants, freelancers, inefficient processes or partial solutions — can be particularly valuable. It demonstrates that a budget exists somewhere around the problem.

You still need to determine whether that spending could realistically move to your solution, but it is stronger evidence than enthusiasm alone.

Negative signals matter too

Startup validation has a natural tendency towards confirmation bias.

You have an idea. You search for evidence that people experience the problem. Unsurprisingly, you find some. But good market research should actively look for reasons the opportunity might be weaker than it first appears.

Perhaps customers complain but don't switch.

Perhaps several competitors have already tried and failed.

Perhaps the problem occurs only occasionally.

Perhaps customers have created free workarounds they are perfectly happy with.

Perhaps the people experiencing the problem aren't the people with purchasing authority.

Perhaps existing products already solve 90% of the problem.

These aren't inconvenient details to remove from the analysis. They're customer signals too.

Individual signals need context

No single Reddit comment, product review, forum discussion or customer interview proves that a market exists. The useful information comes from combining evidence.

You might find:

Direct demand signals — people actively searching for a solution.

Pain signals — complaints about the existing situation.

Adjacent signals — discussions around related problems and processes.

Workaround signals — evidence that customers are already trying to solve the problem themselves.

Competitive signals — what customers like and dislike about existing products.

Behavioural signals — what people actually spend time or money doing.

Willingness-to-pay signals — evidence that the problem already attracts expenditure.

Negative signals — evidence that the problem may not be important enough to support a business.

Individually, each tells you something. Together, they start to tell you what the market actually looks like.

Look for patterns, not quotes

This is particularly important when using online discussions for startup market research. It is remarkably easy to find a quote supporting almost any hypothesis. Search long enough and someone, somewhere, will have complained about exactly the problem your product solves.

That isn't validation.

The more useful question is whether the same underlying behaviour appears independently across different customers, communities and sources.

Are people repeatedly experiencing the same problem?

Are they trying similar workarounds?

Are customers of different competing products making similar complaints?

Are people spending money trying to solve it?

Are the same objections appearing repeatedly?

Patterns are much harder to dismiss than isolated comments.

Customer research is an evidence problem

This is why I think describing startup research simply as "idea validation" can be misleading. An idea doesn't become valid because enough people say they like it.

What you're really trying to do is build an evidence base.

Some evidence will support the opportunity.

Some will weaken it.

Some will be direct.

Some will be adjacent.

Some will come from what customers say.

Some of the most important evidence will come from what they actually do.

The objective shouldn't be to prove that your startup idea is good. It should be to understand the market well enough to decide whether the opportunity is worth pursuing. That's a much harder question. But it's also a much more useful one.