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

Corporate communications audience research should answer a decision that is still open. It should not decorate a message that managers have already decided to send. A sound brief identifies who may be affected, what the organisation does not yet know, how people can participate safely and what evidence would change the communication plan.

This guide uses a fictional case. Northmere Services operates in England and is considering different weekday opening hours at one customer-support site. The authorised research question is narrow: what information would each affected group need before the organisation decides whether, when and how to communicate the proposed change? The research does not decide employment terms, service entitlement or regulatory duties.

No interviews or surveys have been conducted for this article. No findings, quotations, response rates or audience preferences are reported. The sources were checked on 6 September 2026. UK-wide law and guidance are labelled as such; government research guidance is used as a method reference, not as a rule that automatically binds private companies.

Start with populations, not channels

An email list is not an audience definition. Northmere would first create a population register that separates people by their relationship to the decision:

Population	Why it may be affected	What research may establish	What research cannot decide
Employees and worker representatives	Work patterns, responsibilities or access may change	Information gaps, practical barriers and alternative proposals	Contractual rights or whether consultation law applies
Current customers	A route or time for obtaining support may change	Tasks, notice needs and accessible alternatives	Permission to send unrelated marketing
Suppliers	Delivery or escalation contacts may need revision	Which operational hand-offs depend on the current hours	A new commercial obligation outside the contract
Local communities	The site may have a local service or employment role	Questions and effects raised by eligible community participants	Public opinion across England
Investors or lenders	They may need information only if the matter is relevant to their relationship	Which evidence an authorised owner needs to assess materiality	Whether a market disclosure is required
Journalists	They may ask about a public announcement	What factual documents and response routes they need	Editorial interest, coverage or endorsement
Regulators or public bodies	A sector rule or licence may create a notification route	The correct contact, document and evidence format	The regulator's view or legal classification

The buyer, internal approver, participant and eventual message recipient are also different roles. A customer may take part in research but opt out of later contact. A journalist may receive an approved statement without being a research respondent. An employee's manager may commission the work but should not automatically see attributable responses.

Write the decision record before recruiting

The research owner should fill in these fields before approaching anyone:

• proposed decision and options, including retaining current hours;
• English site and eligible population for each stakeholder group;
• known facts, disputed points and evidence owner;
• questions that research can answer;
• decisions reserved for employment, legal, operational or regulatory owners;
• recruitment source, screening rule and exclusions;
• accessible participation routes and adjustments;
• personal data, recordings, transcripts and retention plan;
• analysis method, counter-signals and non-response treatment;
• decision threshold, stop condition and review date.

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 those questions as a demanding ethical benchmark while obtaining advice appropriate to its own facts.

For employees, research cannot be allowed to impersonate consultation. Acas says a workplace consultation should be a genuine two-way discussion in which affected employees can have their views heard. Its guidance on holding a consultation also explains that the process depends on the organisation and issue, and that legal requirements may apply in particular situations. Northmere's qualified employment reviewer must determine whether the proposed change triggers a specific process. A voluntary interview does not waive that route.

Build a frame that can expose omissions

The population register needs a rule for admission, not a list of familiar names. For employees, include people absent from the workplace if the issue affects them. For customers, define 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.

Sampling should deliberately seek variation relevant to the decision, such as use of telephone, web or postal support, working pattern, location and access requirement. It should not assume that a characteristic predicts a view. If protected characteristics or health, disability, trade-union membership or political opinions could be captured, the privacy review becomes more demanding. The ICO's special-category data guidance explains that an Article 6 lawful basis and an Article 9 condition are separate requirements where such data is processed. The page carries a notice that it is under review following data-law changes, so its status needs checking again 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 it cannot establish what all employees believe.

Use interviews to discover meaning

Begin with a small, purposive qualitative stage. Ask participants to describe how they currently obtain or provide support, what happens when the route is unavailable, which facts would help them act and what alternatives they would consider. Ask about a recent relevant incident rather than inviting a general satisfaction score. Follow contradictions instead of smoothing them away.

Question order matters. The Government Analysis Function's questionnaire design guidance notes that related questions should be tested together because order can affect responses. It also recommends qualitative development and cognitive testing. In Northmere's case, a researcher could ask a participant to explain what a draft opening-hours notice means in their own words, then identify missing actions or ambiguous dates. That tests comprehension; it does not test support for the proposed policy.

Every session record should show interviewer, participant category, recruitment route, date, mode, consent status, recording status and any incentive. Notes need separate columns for what the participant said, what the researcher observed and what the team inferred. Verbatim text must not be invented, polished into a stronger view or detached from its context.

Use a survey only for a defined measurement job

A survey becomes useful when the qualitative stage has produced concepts that can be asked consistently across a suitable frame. It is not a cheaper substitute for deciding what a question means. Northmere might measure the proportion of eligible current service users in its sampled frame who report relying on a particular support route during a stated period. The measure would need a precise numerator, denominator, mode, field dates, non-response account and uncertainty.

The questionnaire should avoid forced answers where a person may reasonably prefer not to say. Test navigation, plain language, keyboard use, screen-reader behaviour, alternative formats and completion burden. The Government's inclusive social research guide discusses recruitment, enabling participation and the possible ethical and bias effects of incentives. Again, this is government guidance that can inform a buyer method; it does not certify a private study.

Choose technology after the data journey is approved. A named survey platform's product page is evidence of documented functions, not evidence that Northmere's configuration is accessible, secure or legally appropriate. Record edition, tenant, hosting and transfer terms, user roles, invitation mode, export format, deletion route and supplier changes. Retain a manual interview or buyer-owned form baseline so procurement does not become the research question.

Make privacy and participation real

Participant information should say who is conducting the work, why, what data will be collected, who receives it, how long it is kept, whether participation affects any service or employment decision, how to withdraw and where to raise a concern. The ICO's privacy-information checklist sets out information organisations may need to provide when collecting personal data. The guidance currently notes data-law changes and must be rechecked by a named privacy reviewer.

Collect only fields necessary for the stated question. The ICO's data-minimisation page describes the adequate, relevant and limited test. That means a communications team should challenge a request for names, precise job titles or demographic fields when a coded role and access need would answer the decision.

Employee research has a particular power problem. A person may worry that criticism will affect their work even when an invitation says participation is voluntary. Separate research notes from performance systems, restrict managers' access, report small groups cautiously and offer a representative or independent route where appropriate. The ICO's worker-monitoring guidance stresses transparency, necessity and proportionality in monitoring. A research exercise should not quietly become worker monitoring through identifiers, behavioural logs or linkage.

Accessibility is part of the design rather than a final format conversion. Updated GOV.UK accessible communication guidance recommends involving disabled people and planning alternative formats. Its direct audience is government communicators, but the practical prompt is sound: ask eligible participants what route lets them take part rather than infer needs from a label.

Analyse disagreements without erasing them

Before opening responses, publish a coding plan. Preserve the population and source for each observation. 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 though it is not prevalent in the sample.

Triangulation should be literal. Compare an interview observation with a service record, contract or accessibility test only when the definitions and period align. An operational log can show when calls arrived; it cannot show whether callers understood a notice. A survey response can report a stated preference; it cannot prove future behaviour. Investor, media and regulator records should not be blended into an employee sentiment result.

Use a contradiction table:

Research proposition	Supporting evidence	Counter-evidence	Missing population	Decision consequence
Buyer-authored statement	Evidence ID and date	Evidence ID and date	Group not reached	proceed, revise, hold or narrow

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 offer a useful check against overclaiming a corporate sample.

Turn findings into a bounded decision

The final evidence note should answer only the authorised question. For each stakeholder population, set out the task, evidence heard, unresolved conflict, access need, message implication and owner. Keep a do-nothing option. 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 that part of the decision rather than generalise.

Specify the decision rule before the sponsor sees a preferred answer. For example, pause the communication plan when a material factual dispute remains, when an affected employee group has no valid participation route, when special-category processing lacks documented review, or when the proposed customer alternative has not passed an accessibility test. Reopen research when the 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 all time-sensitive sources on publication day. Research evidence can improve a communication decision, but it cannot create authority, legal permission or stakeholder agreement.