How to run influencer agencies reviews fairly

Influencer agencies reviews should start with an admission decision: is there enough evidence to compare this exact legal entity and service? Public websites can describe an offer, but do not prove creator authority, client experience, campaign control or results.

For an England buyer, a defensible review needs a current supplier dossier, permissioned references, a common test brief and separate pass gates. None was assembled for this desk-researched article, so it names no best agency.

Creatorcampaign.co.uk sets out this reviewing method with an evidence cut-off of 6 September 2026. No agency contact, reference call, software login, creator discussion, contract negotiation or live campaign took place, and there was no supplier funding, affiliate relationship or gifted access.

Named influencer-agency candidates with England operations for the first dossier pass are The Goat Agency, Brave Bison, SocialChain, Whalar, Takumi and Fanbytes. Each candidate is HOLD at this cut-off because no complete verified dossier, permissioned reference or controlled test brief exists for it.

Required public checks per named candidate include Companies House filing date, ICO registration, ASA ruling search, Meta Business Partner directory status and TikTok Marketing Partner directory status. Typical planning range, not a measured public figure: 2 to 4 hours per candidate. Typical dossier validity: 90 days.

A verified dossier has these named sections: company identity, service scope, England availability, creator provenance, rights schedule, disclosure workflow, privacy and PECR map, security controls, pricing units, quote date, two permissioned references and one controlled test brief result. Typical planning range, not a measured public figure: 6 to 10 hours per candidate to assemble the first pass.

Reference rules: each reference must name the client contact, England campaign, service version, dates, permissions in writing and any incentive. Typical planning range, not a measured public figure: 3 to 5 working days to secure two references.

Each named candidate should also be scored against the advertiser-owned in-house baseline on the same units. No named candidate has passed all gates at this cut-off, so no best-agency ranking is issued.

What to take away

A defensible agency review needs a supplier dossier, permissioned references and a common test brief before any comparison.
Desk research can verify public records and supplier statements but cannot report response quality or delivery competence.
Score each mandatory gate pass, hold or fail before any weighted score begins.
Keep agencies, talent managers, creator platforms and in-house teams in separate categories for the same buyer problem.
No agency contact, reference call or live campaign took place for this article, so it names no best agency.

Fix the subject before reading claims

Record the agency's exact registered and trading names, company number where applicable, contracting entity, registered jurisdiction and current filing date. GOV.UK's company information service can show registered information, officers and filed documents. It does not prove service quality, authority to represent creators or suitability for a campaign.

Define the service version as if it were a product: named service, written scope, England availability, channels, creator population, advertiser sector, included deliverables, excluded work, data access, media responsibilities, pricing unit and quote date. "Full service" is not a comparable unit. Neither is a global group website if the proposed contract names a different entity.

Keep agencies, talent managers, creator platforms and in-house teams in separate categories. They may perform overlapping tasks, but a talent mandate is not an advertiser appointment and a platform function is not managed professional advice. Compare candidates only for the same buyer problem.

Publish the reviewer's position

The review record should state:

reviewing organisation, named reviewer and professional status;
exact candidate, service and evidence cut-off;
inclusion and exclusion rules set before evidence collection;
documents opened, version dates and unresolved requests;
reference method and permission to report findings;
synthetic and hands-on tests actually run;
funding, affiliate links, gifts, relationships and other conflicts; and
bounded verdict, expiry and update trigger.

Desk research can verify a live public record and attribute a supplier statement. It cannot report response quality, platform access, creator experience or delivery competence. If hands-on work later occurs, log the test account, permissions, script, sample, expected result, observed result, tester and date. Do not merge that result with a supplier's demonstration.

Use an evidence hierarchy

Different claims need different records. The hierarchy below does not let one strong document compensate for a failed gate.

Evidence class	Appropriate use	What it cannot establish alone
Primary law, regulator or official register	Legal identity, source status and public regulatory context	Complete compliance or service quality
Signed buyer-supplier documents	Exact scope, roles, fees, controls, exit and liability	Delivery before it occurs
Permissioned client reference	That client's described experience under stated conditions	General performance or causation
Controlled buyer test	Behaviour of the tested service or configuration	Untested features or future outcomes
First-party service page	Supplier's current description of its offer	Independent capability or suitability
Case study, testimonial or award	A supplier or awarding body's claim	Reproducible comparative result

The CMA's current online consumer reviews case page records its 2025 guidance and 2026 enforcement activity. A procurement review is not necessarily a consumer-review publication, but the source reinforces why origin, incentive and presentation should be explicit when reviews are used as evidence.

Apply non-compensating gates

Score each gate PASS, HOLD or FAIL. A weighted score begins only after every mandatory gate passes. Buyer preference weights and acceptance thresholds remain blank in this draft.

Identity and authority

Match the contract to the registered entity, bank and insurance evidence. Require written proof of the agency's authority to approach each creator or of the talent manager's mandate. Record conflicts, exclusivity and expiry. A creator profile or agency roster page is a lead, not permission to contract or reuse work.

Creator provenance and safeguarding

Ask how the candidate sourced each creator record, when it was checked, which fields the creator asserts and what platform limits apply. Define eligibility without inferring protected or sensitive traits. If children could perform, route the facts to the safeguarding owner.

GOV.UK describes England rules on child performance licences and supervision; it cannot decide a proposed social post from a supplier pitch.

Advertising recognition and claims

Require a versioned workflow for briefing, claim evidence, creator approval, disclosure, live-render checking, correction and withdrawal. The CMA's guidance for brands assigns practical responsibilities to businesses using creators and intermediaries. CAP's misleading-advertising rules make substantiation a distinct issue from ad recognition. A platform label does not resolve both gates.

Rights and accessible delivery

Request a rights schedule for source files, music, performance, edit, territory, term, organic publication, paid amplification and archive. GOV.UK copyright guidance explains that licences may be limited by purpose, time and place. Payment alone is not evidence of every intended use.

Review captions, audio description, text alternatives, contrast, reading order and accessible landing journeys in the final render. GOV.UK's inclusive communication guidance helps frame audience access needs, but a qualified reviewer must assess the actual formats.

Privacy, PECR and security

Map advertiser, agency, platform and creator purposes separately. Establish controller, joint-controller or processor roles from the facts. ICO controller and processor guidance does not let a contract label override actual decisions.

For outreach, pixels, audience uploads or electronic communications, require a named privacy and PECR review of the real data flow. The ICO's PECR guide is UK regulator guidance and must be rechecked because parts of its data-protection material are being updated after legislative change.

Keep advertiser-owned accounts and assets under named access control. NCSC guidance on protecting social media publishing addresses authorised access, audit trails and stronger sign-in controls. Demand evidence of joiners, movers, leavers, incident response, recovery, logs and revocation in the proposed configuration.

Payments, tax and commercial terms

Separate agency fee, creator cash, commission, gifts or services, paid media, production, expenses, VAT and internal labour. HMRC's online-platform income guidance notes money, gifts and services as information creators may need to consider. It does not determine valuation or treatment in a specific arrangement.

The contract should define payer, payee, evidence, approval, dispute, rebates, credits, cancellation, renewal, correction and exit. A qualified accountant, tax adviser and commercial lawyer must review the actual terms. Never turn an undisclosed quote into a market rate.

Measurement and exit

Require complete metric cards: event, eligible population, numerator, denominator or unit, period, timezone, source and query version, exclusions, missing data, owner and buyer threshold ___. Keep platform-reported observations, attributed commercial events and causal impact apart. HM Treasury's QPIE guidance explains why monitoring a change is not enough to show that an intervention caused it.

Exit evidence covers account ownership, asset and data export, creator communications, payment close, revocation, retention, deletion, ongoing rights, corrections and continuity. Test the route before appointment with synthetic records where possible.

Compare the manual baseline fairly

The baseline is an advertiser-owned register, direct contracting, controlled asset approval, buyer-managed publication evidence and internal reconciliation. It is not free: record buyer labour, specialist review, systems, creator consideration, media, remediation and exit on the same units as each supplier.

The baseline register uses the same units as each supplier: buyer labour hours, specialist review hours, systems cost in GBP, creator consideration in GBP, paid media in GBP, production in GBP, remediation hours and exit hours. Typical planning range, not a measured public figure: 8 to 16 buyer labour hours per campaign for the register.

Apply every gate to the baseline. An internal team does not receive an automatic PASS for data provenance, disclosure, security or safeguarding. Its advantage may be direct ownership; its limitation may be capacity or missing specialist skill. Both are hypotheses until evidenced.

Issue a bounded verdict

HOLD applies when any of identity, exact scope, requested evidence, reference permission or controlled testing is absent. FAIL applies only to a defined mandatory condition the candidate does not meet, with the evidence recorded.

PASS means only that the dossier clears the stated gate for this configuration and date. It is not a compliance guarantee or a prediction of results.

For this publication, the named-provider verdict is HOLD because no complete agency dossier, permissioned reference set or hands-on test exists. The same publication hold applies to the review itself until a named procurement reviewer and specialists verify the source record. Recheck company status, regulator guidance, service documents and conflicts immediately before release.

Qualified review note

This desk-researched method is not legal, privacy, tax, financial, advertising, safeguarding, accessibility, security or procurement advice. Assign named UK reviewers in those disciplines and an evaluation specialist before using the matrix for a real England appointment.

Before you act

Record the exact registered and trading names before reading claims.
Define the service version as a comparable product unit.
Set inclusion and exclusion rules before collecting evidence.
Log any hands-on test with account, permissions, sample and date.
Apply non-compensating gates before any weighted scoring.
Require a rights schedule covering term, territory and paid amplification.

Common questions

What should a review record state about the reviewer's position?

It should name the reviewing organisation, reviewer and professional status, plus the candidate and evidence cut-off. It should also set inclusion and exclusion rules, list documents opened and unresolved requests, and describe the reference method and any tests run.

Why is a platform label not enough for advertising recognition?

CAP's misleading-advertising rules make substantiation a distinct issue from ad recognition, so a platform label does not resolve both gates. A versioned workflow is needed for briefing, claim evidence, creator approval, disclosure, live-render checking, correction and withdrawal.

How should privacy and PECR duties be handled in an influencer campaign?

Map advertiser, agency, platform and creator purposes separately, then establish controller, joint-controller or processor roles from the facts. For outreach, pixels, audience uploads or electronic communications, require a named privacy and PECR review of the real data flow.