When to run audience research for social media agencies

What to take away

Why audience research matters more now

The case for audience research has strengthened because the audience itself has widened. The ONS statistics on internet and social media usage show 96% of adults in Great Britain online in 2024, with social media use at 93%. This is not a niche channel any more.

For a social media agency, that breadth cuts both ways. Almost every client now has an addressable audience online. Assumptions about age, device and behaviour also date quickly. An agency that last refreshed its personas two years ago is working from a faded picture.

Regulation has not stood still either. The ASA and CAP keep publishing research on advertising compliance topics, which matters when advising clients on what can be said in paid and organic posts. The compliance risk sits with the brand, but the agency usually writes the words.

There is a commercial reason too. Clients increasingly ask for evidence. A pitch that shows the sample, the finding and the limits of what is known beats one built on a hunch about which platform is popular.

When should an agency commission audience research?

The trigger points are predictable. They cluster around change.

New client onboarding

A new account is the highest-value moment. You have access to the client's own data, and a finite window to ask questions before delivery pressure takes over. Research at this stage shapes the whole relationship.

Before a strategy refresh

If the last strategy was written before a platform change, a product launch or a shift in the client's customer base, refresh the research first. Strategy built on stale audience data tends to repeat last year's plan with new visuals.

After a performance drop

Falling engagement or rising cost per acquisition often triggers a research brief. That is valid, but it is reactive, and the work usually has to be fast. Speed limits method choice. Keep a lighter version of your standard research ready to run at short notice.

When entering a new market or segment

A client moving into a new region, age group or sector needs fresh evidence. National data can mislead at local level. Northern Ireland, Scotland and Wales each have distinct media habits and public bodies, so assumptions drawn from England do not travel cleanly.

When a client asks for personas

Persona requests often arrive without a research budget attached. Treat them as a research brief in disguise and price the evidence-gathering accordingly. Our guide to building buyer personas for agency clients explains how to turn raw findings into usable profiles.

What does the law require when researching audiences?

The short answer: less than agencies often fear, but more than they often do.

Audience research usually involves personal data, whether survey responses, interview notes or social listening records. UK GDPR and the Data Protection Act 2018 set the conditions. In practice that means a lawful basis, a clear purpose, data minimisation and a retention period.

Social listening deserves particular care. Public posts are still personal data when they relate to an identifiable person. Scraping at scale, or profiling individuals without a lawful basis, creates risk that a short project does not justify.

Consent is one route, but not the only one. Legitimate interests can cover some research activity, as long as a balancing test is documented. The Information Commissioner's Office publishes guidance on lawful bases and on research provisions.

Advertising claims sit under a separate regime. The ASA and CAP codes apply to what clients and agencies publish, including claims drawn from research.

If a campaign says nine out of ten customers prefer a product, the evidence must hold up. The research published by the ASA and CAP shows what the regulators have examined. Use it as a check when a client leans on a statistic.

A practical rule: if you cannot explain in one sentence why you hold a piece of data and how long you will keep it, the process needs fixing before the project starts.

Which research methods fit which question?

Method choice should follow the question, not the agency's favourite tool.

Secondary data is the cheapest starting point. The ONS publishes detailed figures on internet and social media use by age, sex and region, enough to sanity-check most briefs before any primary work begins.

Industry bodies add commercial context. The insight archive maintained by IAB UK covers digital advertising audiences and social media consumption. That helps when a client wants to know how their category behaves rather than the population as a whole.

Primary research fills the gaps. Surveys give breadth and can be quota-controlled to match a target profile. Interviews supply depth and the language customers actually use. If you need volume and spontaneity, listening does that job, though it rarely explains why.

Most agency projects combine two methods: one to size the opportunity and one to explain it. A survey alone tends to produce numbers without reasons. Interviews alone tend to produce vivid quotes that may not generalise.

If you are choosing survey tooling, our comparison of survey tools used by social media agencies covers the trade-offs between panel providers and self-hosted forms.

How should agencies handle social listening?

Social listening is the method most likely to be done badly, because the tools make it feel easy.

Start with a defined question. Asking what people are saying about a brand is not a question, it is a fishing trip. Asking which product complaints appear most often among UK parents discussing weaning is a question.

Then define the sample: platforms, keywords, date range, languages. Record these choices, because they determine what the findings can and cannot support.

Separate observation from inference. A listening report can show that a phrase appears frequently. It cannot show why, unless you ask people. Treat listening as a hypothesis generator, then route the strongest hypotheses into structured customer interviews or a survey.

The social listening training run by the American Marketing Association is a fair benchmark for the skills an analyst needs: query design, sampling logic and interpretation.

Finally, handle storage carefully. Keep raw posts only as long as you need them, anonymise where possible, and avoid building individual profiles from public commentary. The insight rarely depends on knowing who said it.

How do competitors change the research brief?

Audience research is not done in a vacuum. What competitors do shapes what audiences expect, and what audiences expect shapes what your client must deliver.

Competitor context belongs in the brief in three places. In sampling, if a rival dominates a platform, your client's audience may be there because that is where the conversation happens.

Ask which brands respondents considered, not just chose. A rival preference is often a preference for a feature, price point or service promise.

That is why we treat competitor analysis as a companion discipline rather than a separate project. Our guide to competitor research for social media agencies covers how to fold competitor evidence into the same findings document. Clients then see one picture rather than two reports that disagree.

A caution: competitor research is not audience research. Rival activity tells you what is being said. It does not tell you what your client's audience wants. Use it to frame questions, not to answer them.

What does audience research look like before and after?

The table below contrasts a common starting position with a disciplined one. Figures are illustrative examples in pounds sterling, not benchmarks.

| Element | Before | After | | Trigger | Ad hoc, usually after a performance drop | Scheduled at onboarding and each strategy cycle | | Methods | One survey, run by whoever is free | Two methods matched to the question | | Sample | Open link shared on the client's channels | Quota-controlled panel plus recruited interviews | | Social listening | Keyword dump with sentiment score | Question-led query, documented sample, hypotheses routed to interviews | | Legal basis | Not documented | Lawful basis recorded, retention period set | | Personas | Built from team assumptions | Built from findings, reviewed each quarter | | Cost | Absorbed invisibly in account hours | Separate line, for example £3,000 to £8,000 per project | | Output | Slide deck, filed | Findings document with limitations and next steps | | Client involvement | Sign-off at the end | Input at briefing and interpretation | | Review | None | Post-project review of what the research changed |

The cost row matters. When research is absorbed into account hours, it is the first thing cut as a project runs late. A visible line item survives pressure and makes the value easier to defend at renewal.

Which platforms and data sources should you use?

Use platform data where it supports the question, and treat platform documentation as a source rather than a sales pitch. LinkedIn documents its audience targeting capabilities, which helps when a client's buyer is a professional audience and you need to know which targeting dimensions exist.

Population data from the ONS remains the reference point for behaviour across Great Britain. It is not a substitute for primary research, but it prevents the basic error of over-claiming from a small sample.

Sector context from IAB UK is useful in pitches, where a client wants to know whether their instincts match the wider market. Compliance evidence from the ASA and CAP does the same job for claims, and skills training such as the AMA programme gives analysts a route into method rather than tool use.

Four sources, four jobs: population, market, compliance and capability. An agency that uses all four spends less time arguing about whose anecdote is right.

How often should research be refreshed?

There is no single interval, but there are sensible defaults.

Population-level data can be refreshed annually, since the ONS publishes on an annual cycle. Market context can be reviewed twice a year or when a major platform change lands. Personas should be reviewed quarterly in light of campaign performance, and rebuilt annually.

Listening should run continuously for retained clients, with a formal read every quarter. Surveys and interviews are project-based and typically run once or twice a year per client.

The practical test: if a new team member joined tomorrow, could they read your research and know how old it is, what it covers and what it misses? If not, the refresh schedule is the problem, not the research.

Where to start next week

Audience research for social media agencies only earns its place if findings change a decision. Pick one retained client and audit the last piece of research you produced. Check four things: the question it answered, the sample behind it, the legal basis recorded, and the date. Most agencies find at least one gap.

Then fix the process, not just the document. Add research to onboarding, put a review date in the calendar, and price it as a line item. That is the difference between research as a rescue and research as an asset.

Common questions

How much should a small agency budget for audience research?

For a single client project, many agencies work in a range of roughly £2,000 to £8,000 depending on method and sample size. These are illustrative figures in pounds sterling, not market rates. Separate research from delivery hours so it is not silently absorbed. Our cost and budget guide for social media agencies breaks down the components.

Is social listening enough on its own?

No. Listening shows what is said publicly, not why, and it excludes people who do not post. Use it to generate hypotheses, then test them with interviews or a survey.

Do we need consent for every survey response?

Not necessarily. Consent is one lawful basis among several. UK GDPR also allows legitimate interests in some research contexts, provided you document the balancing test. What matters is that a basis is chosen, recorded and explained to respondents in plain language.

Can we reuse client research across accounts?

Rarely, and carefully. Aggregate market data can inform multiple clients. Client-specific findings, especially interview material, should stay with that client. Reusing quotes or personas across accounts risks breaching confidentiality and, where personal data is involved, data protection duties.