Researching influencer-agency buyers with permissioned interviews and evidence profiles

Influencer agencies audience research should begin with the decision a buyer must make. It should not begin with follower demographics or a fictional description of a fashionable brand manager. For an advertiser in England, the useful questions concern authority, campaign risk, evidence and the conditions under which external help is worth investigating.

No interviews, personas or survey findings are reported here. This guide sets out a reproducible method for collecting them. It keeps commissioning organisations, agency suppliers, talent managers, creators, platforms, people exposed to advertising and contracting legal entities in separate populations.

What to take away

Start audience research with the buyer's decision, not follower demographics or fictional brand managers.
Keep commissioning organisations, agencies, creators, platforms and legal entities in separate populations.
Sample roles along the decision path, because budget holders, reviewers and payers may differ.
Recruit authorised adults only; move under-18 questions into a separate study with its own protocol.
Turn findings into evidence profiles with traceable fields, not named personas or invented quotations.
Publish the population, exclusions and achieved coverage beside findings.

Name the decision before recruiting anyone

A clear research decision might be: should this advertiser prequalify an external agency to manage one defined creator campaign, or retain a documented in-house route? That sentence identifies the buyer, choice and service boundary. Add the intended England audience, product category, campaign period and evidence needed for approval.

Do not use "audience" as a catch-all. These groups answer different questions:

Population	What it can evidence	What it cannot evidence alone
Advertiser budget and procurement roles	Buying authority, tender route and commercial constraints	Creator authority or audience response
Campaign, claims and legal reviewers	Approval evidence and reasons to pause	Supplier market demand
Agency delivery staff	Proposed workflow and service boundaries	Independent proof of capability
Talent managers and creators	Mandates, working needs and creator-side process	Advertiser purchase intent
Platforms	Published functions, policies and account controls	Complete campaign coverage or UK legal clearance
People who encounter an advert	Recognition, comprehension or reported response within a study	Buyer satisfaction or agency contribution
Incorporated legal entities	Registered identity and filings	Trading quality, mandates or campaign outcomes

The ASA and CAP influencer-marketing resource collection addresses influencers, brands, agencies and trade bodies. The CMA's brand guidance on social-media endorsements also recognises marketing and talent intermediaries. Those records support the separation of actors. They do not show who buys agency services in England.

Build a frame of commissioning organisations

Start with organisations capable of producing the evidence needed for the decision. The ONS UK business activity, size and location dataset can provide regional, industry and size fields for VAT or PAYE registered organisations. Its quality report documents the annual IDBR snapshot and coverage limits.

Possible England sampling frame, not an influencer-agency buyer list, so set eligibility before contact.

Stratify the frame by English region, industry and organisation size. The ONS dataset covers VAT or PAYE registered businesses, so it cannot identify influencer-agency buyers. Do not treat a broad business count as demand.

An eligible interviewee might work at an English operating site, have inspected a completed creator-campaign contract or approval record in the study period, and hold a stated part in the next supplier decision. Record the number invited, unreachable, screened and eligible. Record interviewed, withdrawn and analysed.

Create a screening log with invited, unreachable, screened, eligible, booked, completed, withdrawn and analysed cases. Record recruitment channel, incentive, access request and reason for exclusion. Sampling through the directory alone over-represents people already aware of the directory.

Within one advertiser, the budget holder, campaign owner, procurement lead, claims reviewer, data or security approver and invoice payer may be different people. Interviewing only the marketing contact can hide vetoes and hand-offs. Sample roles against the decision path, while avoiding duplicate claims from several colleagues about the same commission.

Companies House's company-information service can verify registered details for an incorporated advertiser or supplier. It does not establish that the business bought a campaign, that a respondent has authority, or that a trading name maps neatly to one service team. Confirm those facts from permissioned organisational records.

Require an England connection, a stated role in a recent appointment or rejection, permission to discuss the process and access to at least one relevant record. A recollection without authority may be useful context, but label it accordingly.

Keep recruitment research separate from promotion

Write the invitation around a recent task: reviewing, buying, declining or governing an influencer-agency service. State who is conducting the research, how the person was selected, what participation involves, whether recording is proposed, how the result will be used and when contact details will be deleted.

The ICO's direct marketing guidance distinguishes genuine market research from promotional contact. If the message contains promotion, or the exercise is designed to enable later marketing to participants, direct-marketing rules may apply. A research invitation must not quietly become a sales lead.

GOV.UK guidance on finding research participants recommends defined target groups, clear recruitment criteria, an information sheet, privacy controls and bias checks. Its informed-consent guidance calls for participants to understand the purpose, data collected, observation or recording, sharing, retention, withdrawal and controller identity.

The same guidance asks researchers to include people with access needs rather than designing a route that filters them out.

Research participation and the organisation's UK GDPR lawful basis are not the same decision. Nor does consent to an interview authorise a testimonial, direct marketing or use of a creator's content. A named privacy and PECR reviewer should document the actual roles, lawful basis, privacy information and contact route before recruitment.

Record any incentive in the research file and explain that payment is for time, not a favourable account. Staff participants may face employer policies or power imbalances. Agency or platform funding must be disclosed to participants and readers.

Exclude under-18s until a dedicated protocol exists

Commissioning-buyer research can normally recruit authorised adults. If a question requires under-18 creators or young audience members, move it to a separate study rather than adding them to an adult interview round.

The ICO's updated children and UK GDPR guidance covers child-specific considerations for organisations handling children's information. ASA guidance on advertising and children addresses the separate advertising context. Neither page supplies a complete research protocol.

Before involving a young person, require named safeguarding, research-ethics and privacy reviewers to settle eligibility, age checks, parental or guardian involvement where appropriate, assent, accessible information, incentives, recording, distress routes, withdrawal and deletion. If those decisions are missing, exclude the group and state the resulting evidence gap.

Interview the process, not the respondent's aspirations

Ask a participant to walk through a recent agency decision using records they are allowed to discuss. Useful probes concern the initial brief, creator or talent authority, claim evidence, disclosure approval, access permissions, consideration and invoice separation, incident routes, reporting definitions and exit.

Avoid asking whether influencer marketing is "important" or how much the participant might spend. Those questions invite general attitudes and unverifiable plans. Ask what evidence changed the actual decision, who could stop the work, and what remained unresolved.

Qualitative interviews explore how and why a process worked. They do not estimate prevalence. The 2026 Magenta Book annex on analytical methods notes the depth available from interviews and the risk that willingness to participate introduces bias. Publish the selection route, achieved sample, missing roles and limits on wider inference.

Turn findings into evidence profiles, not fictional personas

A buyer profile should summarise a recurring decision condition found in traceable records. It needs no name, portrait, backstory or invented quotation. Use fields such as decision owned, evidence seen, authority held, campaign context, unresolved gate, counter-evidence and expiry date.

For example, a verified pattern might eventually concern buyers who cannot reconcile creator consideration across a contract and finance record. Until interviews or documents establish that pattern, it remains a hypothesis. Label it as such and specify which evidence would confirm or reject it.

Keep agency staff and advertiser buyers in different profile sets. A talent manager may describe creator-side authority, but cannot stand in for an advertiser's procurement lead. People who recognise an advert can inform disclosure research within a defined study; their responses do not create a supplier persona.

Creator followers and advert recognisers are not proxy buyers. The ASA's 2026 consumer understanding research examined advert recognition within a set participant and platform design. That evidence may inform a separate disclosure question, but it cannot identify which English organisations commission agencies or which approval roles matter. Use a different sampling frame, invitation and analysis for audience members.

Use surveys only for questions a sampling design can answer

A survey can test the frequency of a clearly defined practice if the population, sampling frame and questions are suitable. It cannot repair a convenience frame. Directory visitors, newsletter subscribers and social followers are self-selected groups and should not be labelled English advertisers.

Define the unit before writing questions. One response might represent a legal entity, an English workplace, a campaign decision or an individual role. Prevent multiple colleagues from being counted as separate organisations unless the analysis is expressly person-level. State the reference period and provide "not known" or "not applicable" where the respondent may lack the record.

ONS guidance on uncertainty in surveys distinguishes sampling error from hard-to-quantify non-sampling error. Publish invitations, eligibility, completions, partial responses, item missingness and the calculation behind any percentage. Do not convert a small non-probability survey into an England market estimate.

Tool choice follows the protocol. Record the exact edition, account owner, invitation route, identifiers collected, data location, sub-processors, export, deletion, accessibility evidence and contract. A supplier's security or accessibility statement is evidence to examine, not an assurance for the configured form. Run synthetic tests, then test the final survey with relevant access needs before live recruitment.

Study competitors without pretending to be their customer

Competitor research should record public, dated evidence: registered identity, attributed service descriptions, published policies, ASA rulings, public contract notices and observable directory fields. It must not claim hidden capability, market share, creator relationships or client results.

The ASA rulings database records decisions about particular ads. A ruling is not a score for an entire agency, and absence from search results is not proof that no issue exists. Likewise, a Companies House filing confirms only what the statutory record says.

Do not copy protected portfolio material into a research pack without a rights basis. GOV.UK copyright guidance explains the need for permission, ownership or an applicable exception. Capture the minimum excerpt or structured fact needed, retain the URL and date, and mark supplier statements as claims.

Make the decision trail reviewable

Before findings influence a shortlist, retain the research question, population map, sample frame, recruitment log, information sheet, consent record, discussion guide, codebook, analysis notes, missing-case account and deletion evidence. Separate raw participant data from a redacted decision summary.

The final recommendation should state what the evidence permits. Interviews may reveal approval failures worth testing. A properly designed survey may estimate a defined practice within its sampled population. Public competitor records may identify questions for due diligence. None of these alone proves buyer demand, agency quality, campaign performance or causal impact.

This cluster remains on publication hold until named research, statistics, privacy, PECR, safeguarding, accessibility, advertising, IP and security reviewers approve the configured study. If the population, authority or permission route is unclear, do not recruit.

Publish the boundary

Beside the findings, state the population, frame date, England rule, role criteria, recent-task window, exclusions and achieved coverage. Show missing sectors and approval roles. The result describes recruited decision-makers; it cannot represent all advertisers, creators or audience members in England.

Before you act

Name the buyer, choice and service boundary before recruiting.
Set eligibility criteria and record every recruitment stage.
Separate research invitations from promotional or sales contact.
Document lawful basis, privacy information and contact route.
Exclude under-18s unless a dedicated safeguarding protocol exists.
Interview the process using records, not aspirations or spend plans.
Write the commissioning decision in operational terms.
Separate budget holder, campaign owner, procurement, claims, security and payer roles.
Require an England connection and a recent appointment or rejection.
Log every screening outcome and reason for exclusion.
Keep invitations neutral and exclude participants from sales lists.
Publish the population, exclusions and achieved coverage beside findings.

Common questions

What should a research decision statement contain?

It should identify the buyer, the choice and the service boundary. Add the intended England audience, product category, campaign period and the evidence needed for approval. A clear example is whether an advertiser should prequalify an external agency for one defined creator campaign or retain a documented in-house route.

Why can interviewing only the marketing contact mislead?

Because the budget holder, procurement lead, claims reviewer, security approver and invoice payer may all be separate people, so speaking to marketing alone can conceal vetoes and hand-offs.

What fields should a buyer evidence profile use?

A profile should summarise a recurring decision condition found in traceable records. Use fields such as decision owned, evidence seen, authority held, campaign context, unresolved gate, counter-evidence and expiry date. It needs no name, portrait, backstory or invented quotation, and it should not read as a fictional persona.

Who should be included in influencer-agency buyer research?

Include advertiser-side people with an England connection, a stated role in a recent appointment or rejection, permission to discuss the process and access to a relevant record. Exclude creator followers and advert recognisers, who belong in a separate audience study.