PR agencies audience research: planning buyer interviews

PR agencies audience research should resolve a buyer decision, not decorate a pitch with fictional personas. This guide uses one concrete case: an organisation with an operating establishment in England must decide whether to appoint an external agency for a defined media-relations and issues-preparation package or retain its current in-house route.

For readers using this publication’s PR agency directory, the listings can help identify potential suppliers, but they are not a sample of organisations buying PR or evidence that a provider fits a brief. Examples of established PR agencies a buyer may encounter include Edelman, Brunswick and Grayling; these examples are not recommendations or a claim that any one is listed in the directory.

No research was carried out for this article. There are no interview quotations, survey results, participant numbers or findings to report. The pages cited below describe methods and boundaries. A live study still needs an approved plan, qualified reviewers and evidence from the buyer's population.

What to take away

Name each population before recruiting anyone

The word "audience" can conceal several groups with different roles. Keep them in separate fields.

The ONS business activity methodology distinguishes enterprises from local units and explains that enterprise geography is generally the main operating site or head office. Companies House data products describe legal-entity records. Neither source identifies a communications buyer, a suitable participant or an end audience.

Two further England sources are worth naming. Contracts Finder publishes UK public-sector procurement notices, including contracts awarded by English bodies, and the Charity Commission register lists charities in England and Wales. Both hold organisational records rather than private buying intent.

For the case above, write the eligible organisation population before recruitment. A defensible rule covers legal entities with a verified operating establishment in England and a documented need to decide between the specified external service and an in-house baseline.

They also need a named decision owner within a fixed field period. The population's size stays blank until the frame is built and checked.

Make the buyer segments PR-specific

Do not treat PR buyers as one group. Separate the work being considered, such as consumer media relations, B2B and trade-press work, corporate communications, issues preparation or public affairs, from the organisation’s sector and the participant’s role in the decision.

For an England study, possible sectors to include or exclude deliberately are manufacturers, local authorities, charities, universities and consumer businesses. Within each, distinguish communications or press leads from people who approve budgets, procurement, legal review or delivery. These are proposed sampling dimensions, not findings about the market.

PRCA (Public Relations and Communications Association), CIPR (Chartered Institute of Public Relations) and PRWeek can provide industry and trade-press context for defining terms and questions. They do not establish which organisations are currently buying PR or provide a complete buyer sample.

Turn the decision into research questions

The Government Service Manual's round-planning guidance starts with actionable objectives, problems, assumptions and the decision that evidence must inform. It was written for government services. Here it is adapted as a planning method, not presented as a private-sector rule.

For an agency appointment, useful questions concern events and evidence:

Avoid asking whether someone "values PR" or wants a "strategic partner". Those phrases invite agreeable opinions and do not expose a buying task. Ask for a recent decision, the records used, disagreements, delays and what remains unknown.

Here is a worked England example, illustrative rather than drawn from fieldwork. A manufacturer with a single site in the West Midlands asks whether its in-house team can cover a product launch and a trade-press programme in the next two quarters.

The evidence threshold is a dated launch plan, a named approver, a costed comparison of external and in-house routes, and the record of the last similar launch. A warm view of agencies does not meet it.

Build a recruitment frame that shows its gaps

The Service Manual guide to finding research participants recommends explicit criteria, varied recruitment routes and attention to access needs and recruitment bias. A PR buyer study can begin with authorised customer records, public procurement records or professional networks. PRCA and CIPR can provide industry context, and PRWeek trade-press context, but none is a complete frame of English organisations buying PR. Every route selects a different slice of the population.

Create a row per candidate with source, invitation authority, organisation and establishment rules, relevant decision experience, role, access requirements and exclusion reason. Deduplicate people appearing through several routes, and keep non-response visible.

One completed row, using illustrative placeholders, would look like this:

The row also shows the frame's bias: a trade body list over-represents larger members and misses organisations that buy without joining.

A participant recruited through an agency directory cannot stand for all English organisations. A communications lead cannot answer for finance, procurement or end users unless their role directly supports it.

As a planning range, a single England buyer decision often needs 12 to 20 interviews across two or three strata, with fieldwork over four to six weeks. The number turns on how many distinct strata exist, how risky the decision is and when new accounts stop adding variation. Treat the range as typical guidance, not a rule.

Sampling is purposive when participants are chosen for relevant experience. It can reveal mechanisms and decision steps, but it cannot produce a prevalence estimate without a probability design and suitable response analysis. State the sample achieved by stratum only after fieldwork. Do not write the desired number into a finding.

Conduct interviews without creating evidence in the question

The Service Manual's in-depth interview method recommends a discussion guide, open neutral questions, accessible arrangements and attention to real examples. For this buyer decision, interview the decision owner about a recent appointment or deferral, then speak separately to people who operate approvals, budgets, data and delivery.

Question order matters. Begin with the participant's role and a specific event. Ask them to reconstruct what they did and which record they used. Only then test a possible service boundary. Starting with an agency proposition would prime the account around the proposed answer.

Neutral wordings that avoid priming include:

Record four layers separately:

The Government Service Manual analysis guide distinguishes observations from interpretations and recommends collaborative review. Do not turn one vivid account into a segment or quote it publicly without the agreed permission and context.

Protect respondents and their information

Participation must be voluntary and understandable; the Service Manual's consent guidance says an information sheet should cover who is researching, the purpose, what will happen, data uses and sharing. Covered: retention, observation, recording, withdrawal.

A qualified privacy reviewer must determine the actual UK GDPR basis and documents, despite the government method's usefulness. Participation is not consent to later sales contact, publicity or onward sharing.

Its companion guide on research data and participant privacy recommends collecting only what is needed, controlling access and deleting participant details when no longer required. The ICO's data protection principles remain the regulator reference for lawfulness, fairness, transparency, minimisation, accuracy, storage, security and accountability. ICO guidance may be under review, so recheck its status before fieldwork.

Screeners can expose job, health or other sensitive information. Do not collect protected-characteristic data merely to make a persona look detailed. Define necessity, access, retention and reporting rules first. Never give client observers an identifiable recording by default.

The Service Manual recruitment guidance calls for disabled participants and accessible contact, formats, venues and support. The current GOV.UK Equality Act overview states that the Act applies in England, Scotland and Wales, and the legal position needs a named equality reviewer.

Ask about access needs without requiring unnecessary diagnostic detail. Offer usable information and response formats, allow assistance where chosen, and plan breaks or shorter sessions. Create a safeguarding route for distress, disclosure or power imbalance. A line manager should not observe an employee interview unless the research design and participant choice support it.

Incentives compensate time and access costs; they do not purchase a favourable answer. Record amount, payment route, recruiter interest and any refusal. A respondent should not lose an agreed incentive because they withdraw or criticise the proposal.

Use surveys only for a question they can answer

A questionnaire can collect structured responses from a defined frame. It does not repair biased recruitment, prove causation or make a convenience sample representative. Specify population, sampling frame, invitation method, field dates, questionnaire version, language and access provision before selecting software.

Google Forms, Microsoft Forms and SurveyMonkey have current first-party documentation for response collection and export. Settings differ, and availability can depend on account or plan.

Tool documentation cannot decide lawful basis, PECR, equality duties, security risk or statistical quality. Keep a manual or existing approved collection route in the comparison. Leave a field unknown when the exact edition or configuration has not been verified.

Pilot the wording and accessible route without treating pilot respondents as the main sample unless the protocol allows it. Preserve refusals, partial responses, duplicates and item non-response in the analysis. Report denominators for every proportion only after collection.

Personas are controlled hypotheses

A persona card is useful when it compresses traceable evidence for a decision. It is misleading when a fictional name, photograph and quotation make an assumption look observed. Use a plain hypothesis record instead:

Link every statement to an anonymised evidence reference. Remove a card when later evidence contradicts its eligibility rule or when it no longer changes the appointment decision. Demographics belong only where relevant and lawfully handled.

Research competitors without inventing competence

Companies House records can support legal identity, status and filings, subject to its official disclaimer that published information is not necessarily verified. An agency's own site can support only what it currently claims. Procurement records can show a recorded award and its defined scope. None proves present capacity, quality or client satisfaction.

Create separate columns for register fact, supplier statement, independent primary record and editorial inference. Capture the URL and date, and mark unknowns. Copyright, database, privacy, contractual and site-use questions need their own review.

Finish with a decision, not a persona deck

Before fieldwork, the decision owner should define the evidence threshold for appoint, retain the in-house route, run a limited pilot, or stop. After analysis, publish the frame, recruitment routes, participants by relevant stratum, non-response, questions, field dates, incentive and conflicts.

If an excluded group could change the decision, recruit it or narrow the claim. If consent, access, safeguarding or secure handling fails, pause collection. The useful result may be that the buyer cannot yet justify an agency appointment. That is still a finding, provided the path to it is auditable.

Before you act

Common questions

Why should audience be split into separate populations?

The word audience can conceal several groups with different roles, so the article keeps them in separate fields. Organisation, legal entity, establishment, decision-maker, media contact, consumer and respondent each need their own evidence, such as eligibility rules, contracting authority or consent basis.

What makes a research question useful for an agency appointment?

Useful questions concern events and evidence, such as what happened before external help was considered and who can approve scope and spending. The article warns against asking whether someone values PR or wants a strategic partner, because those phrases invite agreeable opinions.

How should interview data be handled and reported?

The interview section sets out the four layers for handling interview data. Do not turn one vivid account into a segment or quote it publicly without the agreed permission and context.