Audience research for micro influencer marketing

What to take away

The mistake most teams make is starting with the creator list. They open a platform, filter by follower band, and only then ask who those followers are. By the time someone questions the audience, the brief is written and the budget is committed. Research done first changes what you ask creators to make, not just who you ask.

This guide sets out a repeatable method for micro influencer marketing audience research. It covers sizing a niche with official data, mining your own customer records, running surveys and interviews, checking competitors, and scoring the result. It is written for UK teams, and where a rule or dataset covers only part of the country that is stated.

Size the audience before you size the creator

Start with the population you can realistically reach, not the total market. If you sell to new parents in Greater Manchester, the relevant number is not the UK population. It is the number of households in that area with a child under two, adjusted for the share who buy online.

The Office for National Statistics publishes the Population estimates for UK, England and Wales, Scotland and Northern Ireland dataset, found under People, population and community > Population and migration > Population estimates. It covers England, Wales, Scotland and Northern Ireland, broken down by age and region. Use the mid-year estimates to cap your niche at a defensible figure rather than a rounded guess. A niche you can size is a niche you can budget for.

Sizing also tells you when a niche is too small. If your addressable group is a few thousand people, a micro influencer placement may reach most of them, and you can stop worrying about frequency. If it runs to millions, you need a narrower slice before any creator brief makes sense.

Write the number down with its source and date. When a creator's media pack claims a different audience, you now have something to compare against.

Use household income to decide which segments deserve budget

The Department for Work and Pensions publishes the Family Resources Survey. The 2021 to 2022 edition reports household income and spending patterns across the UK. Income bands are a blunt instrument, but they separate segments that behave differently when a creator recommends a £15 product and a £150 one.

Pair income with your own transaction data. If three quarters of your revenue comes from households in one band, that band is your buyer, whatever your brand story says. Segments outside it can still be worth reaching, but they should not absorb the same share of budget.

Be careful with averages. A mean income figure hides the spread, and micro influencers often reach people at the edges of a distribution rather than the middle. Where you can, work with medians and quartiles rather than a single number.

Build personas from evidence, not imagination

A persona is only useful if it changes a decision. If your persona cannot tell you which creator to brief or which claim to avoid, it is decoration. Keep each one to a page and tie every line to a source.

Turn interview notes and transaction data into a one-page profile your creators can actually use in a brief. Keep every line tied to a source, and cut anything that does not change who you brief or what you ask them to say.

Personas should carry the practical details: what the person already owns, what they have tried, what they distrust, and who they follow. Follower counts rarely appear in a good persona. Reasons to buy usually do.

Mine your own data before you buy any

Your order records, support inbox and email list are the cheapest research assets you own. Segment existing customers by product, postcode and repeat purchase rate. Look for the small group that buys twice without a discount, because that group is the one a micro influencer can clone.

If you plan to match customer data or run AI-driven analysis, check the rules first. The UK GDPR and the Data Protection Act 2018 set the underlying requirements, and the Information Commissioner's Office publishes guidance on AI and data protection covering lawful handling of personal data in these workflows.

First-party data also sets your benchmark. When a creator reports a 4 per cent engagement rate, you can compare it with the rate your own email list achieves on the same message. That comparison is worth more than any platform's average.

Run a short survey, then stop

A survey of 100 to 300 people in your niche will not win academic prizes, but it will settle the questions that stall briefs. Ask what they bought in the last six months, which accounts they trust, and what would stop them buying from a creator recommendation.

Keep it under ten questions and offer nothing in return except the results. Completion rates fall sharply once a survey passes a few minutes. If you need a longer instrument, split it into two rounds rather than one long form.

Common tools for a small UK list include SmartSurvey, SurveyMonkey, Google Forms and Typeform. Whichever you choose, handle consent and sampling explicitly when your list is small, and check the UK GDPR and Data Protection Act 2018 requirements before you collect personal data.

For UK readers, the UK GDPR and the Data Protection Act 2018 apply across the whole country, and the ICO regulates them uniformly rather than separately in England, Scotland, Wales and Northern Ireland. Sampling practice, however, often differs by nation because media habits and local press differ.

Talk to five people properly

Five well-run conversations beat fifty survey responses on the questions that matter most: why someone trusted a creator, and what almost stopped them buying. Recruit from your customer list, not from a panel, and pay a small thank-you where that is appropriate.

Ask about the last purchase, not purchase intent. People are unreliable about what they will do and reliable about what they did. A useful opener is: walk me through the last time you bought something because someone online recommended it.

Recruit from your own list where you can, ask about the last purchase before asking about intent, and record findings as themes rather than a transcript pile. Keep the question order consistent so you can compare conversations.

Two cautions. First, five people is enough to find themes, not to estimate proportions. Second, do not let the loudest interviewee become the persona. If one view dominates, look for the contradiction in the next conversation.

Study competitors without copying them

Competitor research answers two questions: which creators already work with brands like yours, and what those brands ask them to say. Both are visible in public posts, and both tell you where the space is crowded.

Log creator-brand pairings over a quarter, note the format, claim and apparent budget tier, then look for the formats that keep recurring. You will not see exact spend, but you can rank activity as low, medium or high by placement frequency and production quality.

The goal is not to copy a rival's creator list, which is likely to be expensive and saturated. It is to find the claims nobody is making. If every competitor in your niche talks about speed, test a claim about after-sales support instead.

Add LinkedIn targeting data where it exists

LinkedIn's audience targeting documentation lists the professional attributes available to advertisers, and those pages double as a map of what the platform knows about its users. That is useful if your niche is defined by job role rather than interest.

Treat platform data as a constraint, not a truth. It tells you what you can filter for, not what your buyer cares about. Use it to shortlist creators, then go back to your interviews to decide the message.

For wider marketing practice, the Chartered Institute of Marketing maintains a content hub of guides and research. It is useful when you need a second opinion on method rather than a platform's own advice.

Decide what good looks like before you look

Write your criteria before you review any creator or segment. Deciding afterwards invites you to bend the rules for a creator you already like. A simple rubric, scored out of five per row, keeps the conversation honest.

Score each candidate segment and each shortlisted creator. Anything scoring below three on two or more rows goes back for more evidence rather than into a live test.

Choosing between two strong segments

When two segments both score well, decide on cost of learning rather than size. The segment you can test for £300 teaches you more per pound than the one that needs £3,000 before it produces a signal.

Also check overlap. If the two segments follow the same creators, you are buying the same attention twice. In that case, run one test and split the creative instead.

Setting a budget you can defend

Use your cost per reach figure, not a platform average, alongside the rubric. For example, a team paying £400 for a placement that reaches 20,000 relevant adults is paying £20 per thousand, which is defensible if a typical customer is worth more than that over a year.

If it is not, the segment is wrong or the price is. Either way, the rubric has done its job before the invoice arrives.

Turn findings into a brief

A brief should carry the persona, the claim, the proof and the one thing the creator must not say. Everything else is optional. Creators who understand the audience produce better work than creators given a shot list.

In the wider campaign plan, sequence tests so each one answers the biggest open question in your research file. One segment at a time, with the persona, claim and proof fixed before the creator brief goes out.

Keep the research file open during the campaign. When a creator's audience responds differently from the persona, that is new evidence, not a failure. Update the persona and brief the next creator with what you learned.

Keep the research alive

Audience research dates quickly. A niche that existed two years ago may have moved platform, changed its language or lost interest in the category. Diarise a light refresh each quarter and a full rebuild once a year.

Reuse the same questions so you can compare answers over time. Changing the wording of a survey question resets your baseline, and you lose the ability to say whether anything shifted.

Finally, keep a one-page summary of what you know, what you assume and what you still need to check. That page is the asset. The survey tool, the persona template and the rubric are all replaceable.

Common questions

How many people do I need for audience research to be useful?

For qualitative interviews, five to eight people in a tight niche will surface most themes. For a survey, aim for at least 100 responses before you quote percentages, and treat anything smaller as directional only.

Can I skip primary research if I already have detailed platform analytics?

Platform analytics tell you what happened, not why. They can size a segment and show which posts worked, but they cannot explain the reason someone trusted a creator, which is the input your brief needs most.

Does the research method differ across the UK?

The UK GDPR and the Data Protection Act 2018 apply across the whole UK, and the ICO regulates them uniformly. Population and income datasets are published for England, Wales, Scotland and Northern Ireland separately as well as together, so use the nation-level figures when your niche is local.

What is the smallest budget that still produces usable findings?

A desk review of official statistics plus five customer interviews can be done for a few hundred pounds if you recruit from your own list. For example, a team spending £250 on incentives and two days of staff time will learn more than one £2,500 placement chosen without research.