Why the broad market label fails, and how to research insurance buyers in five steps

Business insurance audience research should show which firms face a defined decision, how they handle it and what prevents an informed outcome. It should not begin with a made-up persona or treat all small businesses as one buyer.

For an England-focused publisher or service, the core method is straightforward: define the decision, form an exposure-based audience hypothesis, recruit actual or likely users, observe recent behaviour, measure only what a survey can support and keep regulatory product-governance work distinct. This guide sets out that process.

What to take away

Why the broad market label fails

The Department for Business and Trade's Business Population Estimates 2025 estimated 5.0 million private-sector businesses in England at the start of 2025. Its regional table placed 1.042 million in London and 877,000 in the South East.

Those figures are useful population context, with important limits. They include registered businesses and estimates for unregistered businesses, and geography follows head-office location. They do not state how many firms have a particular exposure, hold cover, recently compared a policy or use a specific distribution route.

An England audience therefore needs more detail. A sole trader giving professional advice, an employer operating vans and a retailer leasing premises can sit in the same size band while having quite different questions. Segmenting by employee count or region alone conceals what the insurance service must do.

Step 1: define the decision and audience

Write one sentence describing what the team will decide after the research. Examples include:

Turn assumptions into research questions. "Customers only care about price" becomes "Which fields did recent buyers compare, and what changed their final choice?" "Tradespeople need a mobile journey" becomes "Where and on which device did participants complete each step, and when did they seek help?"

Government Digital Service guidance on planning user research recommends agreeing objectives, identifying groups and selecting methods that provide reliable answers. It says teams can create personas or profiles as they learn, which puts evidence before presentation.

Test the audience hypothesis

Build the hypothesis from characteristics that can change the risk or journey:

Use a statement such as: "Owners or operations leads at small building contractors working in two English regions who obtained or renewed cover after winning a larger contract in the past six months." Add exclusions and explain why they are necessary.

Do not quietly substitute a convenient email list for this group. Existing customers can answer questions about a live service, but they cannot represent eligible firms who did not buy, were declined or chose another route.

Step 2: set boundaries and recruit participants

An editorial audience, advertising segment and insurance product target market are not interchangeable. FCA Handbook PROD 4.2 requires affected manufacturers to identify a target market at sufficient granularity, assess relevant risks, align distribution and consider compatibility and fair value. It also distinguishes the generalised target market from an individual's demands-and-needs assessment at sale.

FCA Handbook PROD 4.3 addresses distribution arrangements and the information distributors obtain from manufacturers. The detailed application depends on the firm and activity. A qualified reviewer should map current rules to the actual model.

Keep a boundary record showing who owns product governance, who conducts customer research, who approves distribution and how evidence moves between them. Audience findings may inform the regulated process, but a publisher's persona is not evidence of compliance by itself.

Recruit people who faced the task

The GDS guide to finding research participants says participants should be actual or likely users. It recommends clear recruitment criteria and inclusion of people with disabilities, limited digital skills or support needs.

Write a screener that confirms business activity, role and recent trigger without revealing the answer sought. Use several recruitment channels where feasible: current users, former users, professional bodies, community groups and a specialist recruiter can reach different parts of the audience. Record the source for every participant.

Avoid recruiting only enthusiastic customers, the team's own contacts or people who repeatedly join research studies. Do not exclude participants because they need a different session format. Ask about access requirements during recruitment and provide an accessible alternative.

GDS notes that interview or usability rounds often use four to eight participants, while surveys and benchmarks can require hundreds for clear findings. These are practice indications, not proof that a given sample represents England. Continue rounds until the decision has enough evidence and important variation has been explored.

Step 3: govern data and interview for behaviour

Insurance research may expose commercially sensitive facts, claim history or personal information. Decide what the study genuinely needs before choosing a recording platform or asking for documents.

GDS guidance on informed consent says participants must understand the purpose, activity, recording, use and sharing of research. Consent should be checked during the session, not treated as a permanent signature obtained at recruitment.

Its guide to managing participant privacy recommends data minimisation, limited access and deletion when details are no longer required. The ICO's research provisions guidance explains the need for a lawful basis and compliance with data-protection principles when processing personal data.

Document controller and processor roles, retention, access, transfers, redaction and withdrawal handling. Do not ask a participant to reveal a policy number, payment detail or unredacted employee record when a blank document or verbal description will answer the question.

Interview for behaviour and context

Ask about one recent journey. Start with the trigger, then reconstruct the sequence: information gathered, people involved, routes tried, quotations received, comparisons made, assistance used and what happened after purchase.

Invite participants to show a redacted document or repeat a task where appropriate. Observe hesitation and workarounds. A confident summary such as "it was easy" can coexist with several calls, spreadsheets and copied answers, so the sequence matters.

Avoid leading questions such as "Would transparent prices build trust?" Ask how the person established cost, what they could not understand and what they did next. Separate a participant's wording from the researcher's interpretation in the notes.

The Financial Ombudsman Service's business protection insurance page identifies complaint themes around cover, notification, settlement, delays and alleged mis-selling. It can help frame topics to investigate. It does not show how prevalent a problem is among all businesses, and interviewers should not presume the participant experienced one.

Step 4: measure and analyse evidence

A survey is useful after qualitative research has clarified the vocabulary and plausible answers. Define the sampling frame, invitation route, eligibility and denominator. Pilot the questions with people matching the intended audience.

Ask about a stated time period and real behaviour. "During your most recent renewal, which documents did you use?" is more measurable than "Do you value clear information?" Include "not sure" where uncertainty is a valid answer, and do not force respondents to select a policy they do not recognise.

Compare survey delivery methods by sample control and accessibility. An invited customer survey has a defined list but excludes non-customers. An open form attracts self-selection. A recruited panel may support quotas but needs auditable screening. Telephone assistance can include people who struggle with a web form but may introduce interviewer effects.

Report question wording, field dates, geography, starts, completions, missing responses, exclusions and weighting. Do not call a convenience sample representative because it is large.

Analyse without manufacturing certainty

Analyse promptly while session context is fresh. GDS guidance on analysing a research session recommends involving the team and separating observations from assumptions as findings develop.

Use an evidence table with the observation, source session, audience characteristics, contrary evidence, confidence and decision affected. Remove unnecessary identifiers. Look for differences as well as recurring patterns: first-time buyers may need vocabulary support, while renewal administrators may struggle to identify changes since the previous declaration.

Personas can summarise validated patterns, but mark small or provisional evidence clearly. Do not invent names, ages, quotations or emotional detail. Retire any persona that no longer changes content, product or support decisions.

Step 5: assess competitor evidence and act on findings

Competitor research can reveal routes and information structures, but it is not customer research. For each insurer, broker or comparison service, record the legal entities, FCA Register entries, insurer panel, advice status, eligibility, documents, price basis, remuneration and after-sale responsibility.

The FCA's guidance for SMEs using insurance brokers highlights questions about advice, limited panels, alternatives, commission and premium finance. Apply the same evidence fields to every organisation and mark anything unavailable.

Never create false personal details or accept a contract to inspect a journey. Public product pages can support a description of stated features; they do not prove service quality or regulatory compliance.

Turn findings into controlled decisions

Each finding should lead to an owner, proposed change, evidence threshold and review date. A content team might replace jargon, expose a policy-document field or add a telephone handoff. A product team might revise screening or test an alternative path. A regulated decision must go through the responsible firm's governance.

Track outcomes connected to the original question: correction rates, successful evidence submission, assisted completion, document retrieval or recurring complaint themes. Page views and lead volume may matter commercially, but they do not show that users understood the cover.

Maintain a decision register containing:

Final checks before findings are used

Confirm that participants match the stated audience, England and UK evidence are not conflated, and population data has not been presented as insurance uptake. Check that every quotation is genuine and consented, every percentage carries a denominator, and limitations sit beside the conclusion they qualify.

Reopen current regulatory and official sources before publication. Keep internal article targets as editorial references until final URLs exist. Most importantly, retain uncomfortable evidence. Audience research earns its value when it changes a weak assumption, not when it decorates a decision the team had already made.

Before you act

Common questions

Why does segmenting by employee count or region fail?

A sole trader giving advice, an employer with vans and a retailer leasing premises can share a size band while facing different questions. Government population figures give context but do not show exposure, cover, recent comparisons or distribution route, so the insurance service's actual task stays hidden.

How should a team recruit research participants?

Recruit actual or likely users, using clear criteria and several channels such as current users, former users, professional bodies, community groups and specialist recruiters. Record each source. Include people with disabilities, limited digital skills or support needs, and ask about access requirements during recruitment.

What separates audience research from FCA product governance?

FCA rules require manufacturers to identify a target market and align distribution, while a publisher's persona is not compliance evidence by itself. Keep a boundary record showing who owns governance, who conducts research, who approves distribution and how evidence moves between them, with a qualified reviewer mapping current rules.