Market research methods: surveys, interviews and secondary data

Market research answers questions that would otherwise be settled by opinion. How many potential customers are there? What do they care about? Would they pay this price? Small businesses can answer many of these questions cheaply, using free official data, a few well-run interviews and a short survey. The skill lies in matching the method to the question and keeping bias out of the answers.

This guide compares three methods: secondary data, interviews and surveys. It covers question design, sample size, bias and the professional standards set by the Market Research Society (MRS).

Which method answers which question

A sensible order for most projects is secondary data first, then interviews, then a survey. Secondary data tells you what is already known. Interviews show you what matters to people and the words they use. A survey then measures how widely those views are held, using questions built from what you heard.

Start with free secondary data

Secondary data is information someone else has already collected. It is quick and often free, and it stops you paying to learn what is already published.

• ONS business statistics. The Office for National Statistics publishes UK business: activity, size and location, which counts VAT or PAYE registered businesses by industry, region and size. Its September 2026 release reports 2.79 million such businesses as of March 2026. The ONS notes that this covers registered businesses only, and points to the government’s Business population estimates for the full picture, including unregistered businesses.
• Nomis. Nomis, a service provided by the ONS, publishes labour market and census statistics at national, regional and local levels. You can use most of the site without registering, and it includes 2021 Census reports for small areas such as wards and towns.
• Company and competitor information. Competitors’ websites, published accounts, reviews and job adverts show prices, positioning and plans.
• Trade bodies and published research. Industry associations often publish summaries of their members’ data.

Check who collected the data, when, how and for what purpose. Note the date and definitions, because “small business” or “customer” may mean something different from your own use.

Use interviews to understand why

Interviews are best early, when you do not yet know the right questions. Speak to 8 to 15 people from your target group as a starting point, and stop when you keep hearing the same things. That number is a rule of thumb, not a statistical requirement.

Ask about past behaviour, not hypothetical future choices. “Tell me about the last time you bought this” yields more reliable answers than “Would you buy this?” Listen for the words people use, the alternatives they considered and what nearly stopped them. Those become your survey options and your marketing language.

Design survey questions that do not lead

Pew Research Center’s guide to writing survey questions notes that “even small wording differences can substantially affect the answers people provide.” Its advice, applied to small business surveys:

• Ask one thing per question. “How satisfied are you with our price and delivery?” is double-barrelled. Split it.
• Use neutral wording. “How much do you love our new range?” invites a positive answer. “How would you rate our new range?” does not.
• Prefer choices to agree or disagree. Pew reports that agree or disagree formats draw extra agreement from some respondents, known as acquiescence bias. Offering a choice between two statements reduces it.
• Think about order. Earlier questions can colour later answers. Put open questions before related closed ones, and randomise lists without a natural order.
• Build closed options from open answers. Pew suggests piloting open questions first, then turning the most common answers into options.
• Pretest. Pew calls pretesting “an essential step”. Try the survey on five people, watch where they hesitate and fix it.

Sample size basics

A survey of a sample estimates what the whole group thinks. The margin of error shows how far that estimate may be from the true figure because of sampling alone. Pew’s explainer on margin of error gives an example: a simple random sample of 1,067 has a margin of error of about plus or minus 3 percentage points at 95% confidence.

For a simple random sample and a result near 50%, the margin of error is roughly 1.96 times the square root of 0.25 divided by the sample size:

Three cautions follow. Quadrupling the sample only halves the margin. Subgroups have larger margins because they contain fewer people. And, as Pew stresses, the margin covers sampling error only. It says nothing about who was left out, how questions were worded or whether people answered honestly. A survey sent only to your newsletter list tells you about your newsletter list, not the whole market.

An example (illustration only)

A tutoring business in Pune is considering a new online course for parents in UK cities. It first reviews published data on its target areas, then interviews 12 parents from its existing network about how they chose tutoring in the past. Parents keep mentioning trial lessons and flexible timing. The business builds a survey around those themes and gets 400 responses through partner schools. About 60% say a free trial would matter in their choice. With a margin of roughly plus or minus 5 points, the true figure among parents reached this way is likely between about 55% and 65%. Because the sample came through partner schools, the business treats the result as a guide for those families, not all UK parents. The business and figures are invented to show the reasoning.

Bias to watch for

• Selection bias. Only reaching people who already like you.
• Non-response bias. The people who answer differ from those who do not.
• Leading questions. Wording that suggests the answer you want.
• Social desirability. People giving the answer that sounds better, such as saying they would pay more for sustainable products.
• Confirmation bias. Noticing only the answers that support your plan.

Research ethics and the MRS Code of Conduct

The Market Research Society’s Code of Conduct is the UK profession’s main standard. The current version came into effect on 14 August 2023 and binds MRS members and Company Partners. Among its changes, it gives more weight to participant wellbeing and requires reports to describe how samples were defined, including what makes them “representative”.

Even if you are not a member, its principles are good practice. The MRS guidelines on questionnaire design describe research as depending on the “willing co-operation” of the public and businesses, and state that its purpose is to “collect and analyse information and not to create sales”. In practice: tell people who you are and why you are asking, keep surveys short, let people skip questions, protect their data under UK GDPR, and never dress up a sales pitch as research.

Common mistakes

• Running a survey before you know which questions matter.
• Treating interview findings as percentages.
• Surveying friends, family or loyal customers and calling it the market.
• Asking people to predict what they would pay instead of studying what they have paid.
• Ignoring free official data that already answers the question.