Market research is commissioned for one of two reasons: to inform a decision that has not been made, or to justify one that has. Only the first produces anything useful.
Before starting, write down what result would change your mind. If nothing would, you are not doing research.
Size the market from the bottom up
Top-down sizing — take an industry report figure, assume a share — is the most common approach and the least credible. Bottom-up forces you to state assumptions that can be challenged, which is exactly what makes it persuasive.
- Define precisely who the customer is. Not "small businesses" but "UK dental practices with two to ten chairs".
- Count them, from a source you can cite — a trade register, company database or government statistic.
- Estimate what each would pay, from your own pricing or observed comparable pricing.
- Multiply, then apply a realistic reachable proportion given your channel.
- Present a range with the assumptions listed, not a single number.
TAM, SAM and SOM are only useful if each is derived rather than assumed. A SOM that is simply "1% of TAM" tells a reader nothing except that the work was not done.
Sources you can actually cite
- Government statistics — census, labour force surveys, trade data. Free and authoritative.
- Regulator and industry body publications — often contain figures paid reports simply resell.
- Company filings — annual reports of listed competitors reveal pricing, margin and segment sizes.
- Academic literature — particularly for behaviour and adoption research.
- Trade press — useful for direction and timing, weaker for hard numbers.
- Your own primary research — interviews and surveys, which nobody else has.
Record the date on every figure. A 2019 market size quoted in 2026 without comment undermines everything around it.
Customer research: watch behaviour, not intent
People are poor predictors of their own future behaviour and reliably polite about new ideas. "Would you use this?" produces enthusiasm that never converts.
- Ask about the last time they had the problem, in detail. Past behaviour is observable.
- Ask what they currently do instead, and what it costs them in time or money.
- Ask what they have already paid for to solve it. Spending is the strongest signal there is.
- Avoid describing your solution until the end — it biases everything before it.
- Watch for the difference between "that's interesting" and "when can I have it".
Map competition honestly
The real competitive set is wider than the companies that look like you. It includes substitutes and inertia.
- Direct — same solution, same customer.
- Indirect — different solution to the same problem.
- Substitute — the spreadsheet, the agency, the manual process.
- Doing nothing — usually the market leader, and the hardest to displace.
Report it so it can be used
- Lead with the answer to the question that prompted the research.
- State confidence honestly — which findings are solid, which are indicative.
- Show the working for anything quantitative.
- Include the evidence that contradicts the conclusion. Its absence is what a sceptical reader looks for.
- End with the decision the research supports, and what would change it.
Questions this raises
For exploratory work, patterns usually stabilise between fifteen and twenty-five within a single segment. If every new interview still surprises you, keep going.
Occasionally, for hard-to-find segment data. Check first whether the underlying figures come from a public source you can cite directly.
Still stuck after reading this? That is usually the point at which it is worth asking someone. Describe your project or ask on WhatsApp.
