Audience research: how UK teams brief content agencies

What to take away

Defining audience research for content work

Audience research answers a narrow question: who are we writing for, what do they need, and how will we know? It is not the same as market sizing or brand tracking, though the three often share data.

For a content team, the output is a set of decisions. Which topics to cover. Which formats to produce. Which channels to publish on. Which calls to action to test.

A useful working definition is that audience research converts scattered signals into a documented view of the reader, their context and their language. That view is then reviewed as evidence changes.

It differs from persona writing, which is a presentation layer. Personas are one way to communicate findings; they are not the research itself.

Why audience research matters commercially for content agencies

Commissioning content without audience evidence tends to produce material that satisfies internal stakeholders and misses the reader. The cost shows up later as low engagement and rework.

Agencies that research first can price and scope more accurately, because the brief is grounded in a defined audience rather than an open-ended topic list. Clients get a defensible rationale for what was made and what was left out.

Research also reduces risk. If a campaign underperforms, evidence shows whether the audience definition, the message or the channel was the weak point. Without it, teams argue about opinions.

There is a regulatory dimension too. Claims about what an audience thinks or does need support, and the ASA and CAP research pages explain how evidence informs advertising standards work in the UK.

What good audience research contains

A complete programme usually has five layers. They can be delivered in stages, but each layer should be traceable to a decision.

Desk research and published data

Desk research establishes the baseline before any primary work begins. It covers market reports, sector statistics, search demand data and government datasets.

For UK digital audiences, the ONS internet access dataset is a sensible starting point because it covers device use, access and frequency of use across household types. It helps a team avoid assumptions about who is online and how.

Desk research is cheap and fast. Its weakness is that published data is rarely specific to one niche, so it sets context rather than answers.

First-party data and behavioural evidence

First-party sources include analytics, search console data, CRM records, support tickets and sales call notes. These show what real visitors and customers actually do.

Behavioural evidence is often more useful than stated preference. A reader who says they want long-form guides but abandons pages after twenty seconds is telling you something different.

Segment first-party data carefully. Small samples produce unstable patterns, and mixing anonymous analytics with identified customer records raises privacy questions that should be resolved before analysis, not after.

Primary qualitative research

Interviews, focus groups and diary studies explain motivation. They are the right tool when you need language, objections or context rather than prevalence.

Sample sizes are small by design, so qualitative findings should be reported as themes, not percentages. A handful of interviews can reshape a messaging strategy without proving how common a view is.

Recruit deliberately. Convenience samples of existing customers will flatter the brand and miss people who considered it and chose elsewhere.

Quantitative surveys and testing

Surveys measure how widespread a view is. They work best when the question set is short, the sample is defined and the analysis plan is agreed before fieldwork starts.

Message testing, headline testing and format testing all sit here. Each produces evidence you can act on within a production cycle rather than a research report that arrives too late.

Platforms such as SurveyMonkey, Typeform and Google Forms can support surveys; message and headline testing may use different workflows. For a comparison of survey platforms and question-design trade-offs, see the guide to survey tools for content agencies in England.

Segmentation models compared

Segmentation is where most audience research either becomes useful or stalls. Common models include demographic segmentation (grouping by personal characteristics), firmographic segmentation (grouping organisations by characteristics such as sector or size), behavioural segmentation (grouping by actions), needs-based segmentation (grouping by needs), psychographic segmentation (grouping by attitudes or values) and jobs-to-be-done (grouping by the progress people seek). The table below sets out common models and what each is good for.

Most content programmes end up with a primary model and one supporting lens. Adding more lenses rarely improves decisions and usually slows production.

If your organisation sells to other businesses, the firmographic and needs-based models often combine well. Use a clear definition of your target audience in England to narrow the scope before briefing an agency.

Accessibility and inclusion in audience research

Accessibility is both a research subject and a research method. If your recruitment excludes people with access needs, your findings will exclude them too.

GOV.UK research into delays in decision-making for reasonable adjustments shows how access needs interact with process and communication. Content teams can apply the same lesson: ask how people access information, not just whether they do.

Practical steps include offering multiple participation formats, allowing longer response windows, testing documents with screen readers, and publishing findings in accessible formats.

The commercial case is straightforward. Formats that work for people with access needs, such as clear structure and captions, tend to work better for everyone.

Privacy, consent and data handling

Audience research collects personal information whenever data can be linked to an identifiable person, even indirectly. The ICO definition is the reference point, within the framework of the UK GDPR and the Data Protection Act 2018.

In practice, that means agreeing lawful bases before fieldwork, minimising what you collect, setting retention periods and documenting how data flows between client and agency.

Where an agency processes personal data on the client’s behalf, agree the controller and processor roles and put a data processing agreement in place where required. Also agree the privacy notice wording, retention schedule, access controls, and whether data must be returned or deleted when the project ends.

Consent is one route, not the only one. Legitimate interests may cover some analytics and research work, but the assessment must be recorded and specific.

Anonymisation is harder than it looks. Combining a job title, employer size and region can identify a person in a small market, so check before publishing segment detail.

Research methods and their trade-offs

Choosing a method is a trade-off between speed, cost, confidence and depth. No single method covers everything, so pair a broad method with a deep one.

A common pattern is a quarterly survey plus six to ten interviews, with continuous analytics review. That mix keeps evidence current without a heavy annual project.

The sequence matters more than the tooling. Desk research, then qualitative themes, then quantitative confirmation, then testing in production.

How to brief an agency on audience research

A good brief states the decision, the deadline and the constraints. Agencies cannot scope research without knowing what will change once findings arrive.

Include existing evidence: analytics access, previous reports, customer lists and any known gaps. Name the internal owner who will act on the findings.

Agree deliverables in advance. A slide deck, a segment definition document and a messaging framework are different products with different effort levels.

Set out governance early: who approves recruitment, how incentives are handled, where data is stored and what happens to it at the end of the project.

Audience research brief checklist

For buyer persona work, the guide to buyer personas for content agencies in England covers the artefacts and review cycle in more detail.

Shortlisting and appointing an agency

Use directory listings to identify potential content agencies, then check directly how listing inclusion is decided and whether each agency has relevant audience-research experience. A directory listing alone does not establish that an agency is the right fit for a particular brief.

Give shortlisted agencies the same brief so their proposals can be compared on scope and assumptions. Assess whether their recommended methods fit the decision, whether their recruitment plan reaches the intended audience, and whether they explain sample limitations and quality checks. Also compare the proposed team, deliverables, timetable, data-handling arrangements and itemised costs. Ask for relevant examples of previous work and clarify who is accountable for delivery.

Turning findings into content decisions

Research that does not change production is wasted. Map each finding to a decision, an owner and a review date.

Translate segments into editorial rules: which topics each segment gets, which formats, which channels and which proof points. Keep the rules short enough to be used.

Build a message bank from real customer language. Direct quotes and objections are more useful to writers than abstract attribute lists.

Review the model quarterly. Audiences shift with technology, regulation and economic conditions, so a persona set built three years ago is a historical document.

Competitive context belongs here as well. Knowing which questions rivals answer well helps you avoid duplicate content and find gaps; the guide to competitor research for content agencies in England explains how to investigate this.

Governance, ethics and quality control

Set minimum standards for sample size, recruitment and reporting. Document them so different researchers produce comparable work.

Ask the agency to state the proposed sample, recruitment source, inclusion and exclusion criteria, and why that design answers the brief. There is no universal minimum sample size: survey adequacy depends on the population and design, while qualitative recruitment should be justified by the question and depth of evidence. Require limitations to be reported, and do not present a small purposive sample as representative.

Separate findings from recommendations. Findings describe what the evidence shows; recommendations describe what the team proposes to do about it.

Record limitations honestly. Response bias, small samples and tracking gaps all belong in the report, not in a footnote nobody reads.

Keep an evidence log with dates and sources. It makes future updates faster and protects the team when decisions are questioned.

Ask whether the agency follows the MRS Code of Conduct and how it applies the code to recruitment, participant information, incentives, confidentiality and reporting. Request details of the proposed quality checks and any relevant limitations; a code does not replace a clear method or appropriate data-protection arrangements.

What this costs and how long it takes

Timelines depend on recruitment and review cycles rather than analysis. Desk research can take a week; a survey with fieldwork typically runs three to five weeks.

Budget varies with sample size and method. For example, a mid-sized UK content programme might allocate £6,000 to £18,000 for a combined desk, interview and survey phase, though that is an illustrative range rather than a market rate. Ask agencies for itemised proposals against the same brief, including recruitment, incentives, fieldwork, analysis and deliverables, and compare the assumptions as well as the totals.

Ongoing costs are lower. Quarterly refreshes and continuous analytics review usually need days rather than weeks.

The bigger cost is internal time: briefings, recruitment support, review meetings and the production changes that follow.

Common questions

How much audience research does a content strategy need?

Enough to make the main decisions defensible: audience definition, priority topics, formats and channels. For many UK teams that means desk research, a small interview round and one survey per year.

Can an agency run research without accessing customer data?

Yes. Desk research, published datasets and recruited interviews can carry a programme. First-party data improves precision but is not a prerequisite, and it brings privacy duties.

How often should audience segments be reviewed?

At least annually, and sooner if your market, product or regulation changes. Treat the segment document as a living asset with named owners.

What makes audience research evidence weak?

Convenience samples, leading questions, no documented method and findings that cannot be traced to a decision. Weak evidence is usually a design problem, not a data problem.