Market research without a budget: five methods that work

Most market research fails not because the budget was small but because the question was unanswerable. You do not need money to learn what you need; you need a question that a cheap method can settle.

Start with the question, not the method

Three questions are answerable cheaply and cover most early decisions: does this problem exist for enough people, what do they currently do about it, and would they change. A survey that asks all three badly is worth less than five conversations that establish the first one.

The questions that are not answerable cheaply: how much would they pay (a stated intention), would they buy this (also a stated intention), and what is the size of the market (a number that requires paid data). Treating these as researchable is where budgets disappear.

Method one: conversations about the last occasion

The highest-value method and the cheapest. Ask about the last time the person had the problem, and walk through what happened chronologically. What prompted it, what they did first, what they tried next, what finally worked, what it cost in time and money.

The reason this works is that it asks about behaviour that has already happened. Hypothetical questions produce polite answers; a reconstruction of an actual event produces facts.

Method two: reading what people already wrote

Reviews of competitors, support forums, community threads and complaint sites contain more useful language than any interview you will run, and they are free. The value is partly content and partly vocabulary: the words customers use for the problem are the words your copy should use.

Read for the failure mode rather than the verdict. A two-star review explains what went wrong; a five-star review mostly confirms the category was acceptable.

Method three: watching what people do

Behaviour that leaves a trace beats behaviour that is reported. Where do people abandon a signup, what do they search for after landing on the page, which question does support actually receive? Every one of these is recorded already and almost none of it is read.

Method four: a small test

A page with two versions of the same message, a small audience, and one action will settle a question that no amount of opinion can. The discipline is to test the thing most likely to be wrong: not the button colour, but whether anyone wants the outcome at all.

The trap is that a test with a small sample produces a number for every version, and a number looks like a result. Decide in advance what difference would count as a difference, and treat a result below your sample size as no result.

Method five: asking the people who said no

The cheapest research nobody does. Ask people who did not buy why, and ask people who left why. The answers are more accurate than anything a marketing survey produces, because the person has no reason to be encouraging and has already resolved the question.

The method for turning conversations into an audience definition is in defining your audience. For measuring whether the result changed anything, see measuring marketing honestly.

Why cheap research is usually done badly

The failure is almost always the same: the research is designed to confirm a decision already made. The questions are leading, the sample is people who already agree, and the conclusion is the plan that existed beforehand. This is more expensive than no research, because it produces false confidence.

The countermeasure is to write down, before you start, what result would change your mind. If no realistic result would, the research is theatre and should be skipped in favour of shipping.

Recording what you learn

Cheap methods produce unsearchable material: recordings, notes, threads. Keep a single file of verbatim quotes with the date and the source, because the sentence a customer actually used is worth more than your summary of it, and summaries drift toward what the team already believes.

Review that file before writing any copy. The purpose of research is not a report; it is a vocabulary and a list of constraints you would not otherwise have known.

When to stop researching

Research has diminishing returns and a clear stopping point: when the next conversation is unlikely to change a decision. In practice that is often after five to ten interviews, well before a sample large enough to satisfy anyone who wants a number.

The alternative to more research is a small reversible decision. Where the cost of being wrong is low and the change is easy to undo, ship and learn from what happens. Where the change is expensive, research first. Choosing between the two based on the cost of being wrong is more useful than a general preference for rigour.