Audience research for corporate communications that starts with populations, not channels

Audience research begins while a communication decision remains open, so evidence can alter the plan instead of lending authority to a settled message. A decision brief should name affected people, unknowns, safe participation routes and the evidence that could change the communication plan.

The fictional Northmere Services case concerns a proposed change to weekday opening hours at one customer-support site in England. The authorised question is what information each affected group needs before deciding whether, when and how to communicate the proposal; research cannot decide employment terms, service entitlement or regulatory duties.

Sources were checked on 6 September 2026. UK-wide law and guidance are labelled as such, and government research guidance is a method reference rather than a rule that automatically binds private companies.

If managers have settled the message, reopen or stop the decision

If managers have already settled the message, pause or reopen the communication decision before commissioning research. Audience work should answer an unresolved question and name evidence that could change the plan, rather than make a settled choice appear to have audience support or disguise who has authority to decide.

For Northmere, the authorised question concerns what information each affected group needs before the organisation decides whether, when and how to communicate the proposed opening-hours change. The research must not decide employment terms, service entitlement or regulatory duties; those boundaries keep the work separate from decisions it has no authority to settle.

• Rule of thumb: Research only while evidence can still change the communication decision.

If you know the channel but not the people, register populations

If an email list or another familiar route is being treated as the audience, replace it with a register of people affected by the decision. Different populations can experience different effects, and research may establish different things for each; the route used to reach them is not the population definition.

Keep the buyer, internal approver, research participant and eventual message recipient distinct. Buyer personas can help identify these different roles without treating them as interchangeable. A customer can take part in research and decline later contact; a journalist can receive an approved statement without becoming a respondent. An employee's manager may commission the work, but should not automatically see responses that identify participants.

• Rule of thumb: Define people by their relationship to the decision, not the route used to contact them.

If employees may confuse research with consultation, set the boundary

If employees are affected, establish whether a specific consultation process applies before inviting research participation. A voluntary interview does not replace that process, and taking part does not waive a route that may be legally required. Northmere's qualified employment reviewer must determine whether the proposed change triggers a specific process.

• Record personal data, recordings, transcripts and the retention plan before approaching participants.

The Government Social Research profession's ethical assurance guidance names specific informed consent, enabling participation, sound interpretation and minimising harm among its principles. Its direct scope is government research. A private employer can use these questions as a demanding ethical benchmark while obtaining advice that fits its own circumstances.

Acas says workplace consultation should be a genuine two-way discussion in which affected employees can have their views heard. Its guidance on holding a consultation explains that the process depends on the organisation and issue, and that legal requirements can apply in particular situations. Research participation itself cannot determine whether those requirements apply.

• Rule of thumb: Research participation never settles whether workplace consultation is required.

If familiar contacts define the sample, write admission rules

If a population register relies on familiar names or easy contacts, add a clear eligibility rule for every group.

A sampling frame that exposes omissions is more useful than one that quietly leaves less visible people out. State who qualifies before invitations go out, rather than interpreting a convenient sample as the whole population.

For employees, include people absent from the workplace if the issue affects them. For customers, specify the service and period that make someone eligible. For suppliers, identify contracts with a genuine dependency. Community recruitment needs a declared geographic boundary and a route beyond the organisation's closest contacts.

Seek variation relevant to the decision, including use of telephone, web or postal support, working pattern, location and access requirement. Do not assume that a characteristic predicts a view. If protected characteristics, health, disability, trade-union membership or political opinions could be captured, the privacy review needs greater care.

The ICO's special-category data guidance explains that processing such data requires an Article 6 lawful basis and a separate Article 9 condition. The guidance page is under review after data-law changes, so check its status before fieldwork.

Recruitment logs should record invitations, eligibility, refusals, withdrawals, unreachable contacts and access adjustments. Do not replace missing groups with a confident narrative. A voluntary sample of readily available colleagues may reveal wording problems, but cannot establish what all employees believe.

• Rule of thumb: Make eligibility and missing participation visible before interpreting responses.

If participation could expose people, build in privacy and access

If collecting personal information or making participants identifiable is possible, build privacy and access controls into the research before fieldwork. Accessibility and protection are design requirements, not post-production fixes; collect only fields needed for the authorised question and make participation routes safe for each group.

Participant information should say who runs the research and why, what data is collected, who receives it and how long it is kept. It should explain whether taking part affects service or employment decisions, how to withdraw and where to raise a concern. These details help participants understand the terms before they decide whether to take part.

The ICO's privacy-information checklist describes information organisations may need to provide when collecting personal data. Its guidance notes data-law changes, so a named privacy reviewer must recheck its status. Collect only fields needed for the stated question, applying the ICO's adequate, relevant and limited test.

Challenge requests for names, precise job titles or demographic fields when a coded role and access need would answer the decision. Employee research has a power problem: people may fear that criticism will affect work even when invitations say participation is voluntary. Design the route so that fear does not quietly shape who responds.

Separate research notes from performance systems, restrict managers' access, report small groups cautiously and offer representative or independent routes where appropriate. The ICO's worker-monitoring guidance stresses transparency, necessity and proportionality; identifiers, behavioural logs or linkage must not quietly turn research into worker monitoring.

Updated GOV.UK accessible communication guidance recommends involving disabled people and planning alternative formats. Its direct audience is government communicators, but the practical prompt also applies here: ask eligible participants which route lets them take part rather than inferring needs from a label.

• Rule of thumb: Collect only what the decision needs, and make participation safe and accessible.

If people do not understand the issue in the same way, interview first

If the team does not yet know how affected people understand the issue, begin with a small, purposive interview stage.

Use interviews to discover meaning before deciding whether a consistent survey question can be asked across a suitable frame. Ask about a recent relevant incident rather than beginning with a general satisfaction score.

Ask participants how they currently obtain or provide support, what happens when that route is unavailable, which facts would help them act and what alternatives they would consider. Follow contradictions rather than smoothing them away. In Northmere's case, a researcher could ask someone to explain a draft opening-hours notice in their own words, then identify missing actions or ambiguous dates.

Question order matters. The Government Analysis Function's questionnaire design guidance says related questions should be tested together because order can affect responses; it recommends qualitative development and cognitive testing. Asking someone to explain a draft notice tests comprehension, not support for the proposed policy.

Each session record should include interviewer, participant category, recruitment route, date, mode, consent status, recording status and any incentive. Keep participant statements, researcher observations and team inferences in separate note columns so that the evidence retains its source and context.

Do not invent verbatim text, polish notes into a stronger view or detach them from their context. Those safeguards make it possible to distinguish what was said, what was observed and what the research team inferred, without merging the three into one claim.

• Rule of thumb: Use interviews to learn what a question means before measuring answers to it.

If answers can be measured consistently, specify the survey job

If qualitative work has produced concepts that can be asked consistently across a suitable frame, a survey tool can serve a defined measurement job. It is not a cheaper substitute for deciding what a question means. Set the measure and its limits before choosing technology or inviting responses.

Northmere might measure the share of eligible current service users in its sampled frame who report relying on a particular support route during a stated period. Specify the precise numerator, denominator, mode and field dates. The measure also needs an account of non-response and uncertainty; it cannot establish future behaviour from stated answers.

For a questionnaire, avoid forcing an answer where someone prefers not to say. Test navigation, plain language, keyboard use and screen-reader behaviour, as well as alternative formats and completion burden. The Government's inclusive social research guide addresses recruitment, enabling participation and possible ethical and bias effects of incentives.

That guide can inform a buyer's method, but does not certify a private study. Choose technology only after approving the data journey: a survey platform's product page documents functions, but does not prove that Northmere's configuration is accessible, secure or legally appropriate.

Record edition, tenant, hosting, transfer terms and user roles, along with invitation mode, export format, deletion route and supplier changes. Keep a manual interview or buyer-owned form as a baseline so that procurement does not become the research question.

• Rule of thumb: Choose a survey only after its measure, frame and data journey are defined.

If evidence conflicts, preserve each source and its limits

If accounts conflict, preserve the disagreement and its source instead of reducing it to one confident narrative. Before opening responses, publish a coding plan and retain the population and source for each observation. Do not treat disagreement as noise to be removed from the decision record.

Report how many eligible people were approached, participated, declined, withdrew or did not respond, without exposing identities in small groups. Separate recurring themes from isolated but high-consequence barriers. A rare accessibility failure can still be decisive even when it is not prevalent in the sample.

Triangulate only when definitions and periods align. An operational log may show when calls arrived, but cannot show whether callers understood a notice. A survey response can report a stated preference, but cannot prove future behaviour. Do not blend investor, media and regulator records into an employee sentiment result.

The UK Statistics Authority's standards for official statistics require official producers to use suitable sources and explain quality, limitations and uncertainty. Northmere is not producing official statistics, but those disciplines provide a useful check against overclaiming from a corporate sample.

• Rule of thumb: Keep contradictions, gaps and uncertainty visible in the evidence note.

If a material gap could change the outcome, hold or narrow the decision

If a material factual dispute remains, pause the communication plan. The final evidence note should answer only the authorised question and keep a do-nothing option available, rather than stretch findings beyond the populations or issues the research covered.

Missing evidence is a reason to narrow or hold a decision, not to fill gaps with assumptions.

For each stakeholder population, set out the task, evidence heard, unresolved conflict and access need, then add the message implication and owner. If customers can use current hours without evidenced difficulty, the case for change may weaken. If absent employees or users of an accessible route were not recruited, hold the decision rather than generalise.

Set the decision rule before the sponsor sees a preferred answer. Pause if an affected employee group lacks a valid participation route, special-category processing lacks documented review, or the proposed customer alternative fails an accessibility test. Reopen research if service design, population, channel or purpose changes.

Publication remains on hold. A named UK research lead must approve the method; employment, privacy, accessibility, security and corporate reviewers must clear their own boundaries. A fact-checker must reopen time-sensitive sources on publication day. Research can improve a communication decision, but cannot create authority, legal permission or stakeholder agreement.

• Rule of thumb: Hold the decision whenever a material gap could change who is affected or what they need.

If fieldwork is imminent, complete the final checks

If recruitment is next, hold invitations until the decision record, boundaries and participant safeguards are complete.

The final check catches missing approvals and stops a convenient contact list, an untested question or an unchecked source from becoming the basis for the research. Begin only when the method can answer the authorised question safely.