How to judge creator discovery platforms benchmark research

This method is for teams judging creator discovery platforms benchmark research: it sets out how to test whether an external figure — from a platform such as Modash, HypeAuditor or CreatorIQ, from supplier analysis or from published research — may be admitted to a comparison. Apply the same fields and the same record to whichever supplier report, dashboard export or published study you hold: the test decides admission, not the brand on the cover.

There is no official England benchmark for creator-platform discovery, contact, response, publication or commercial effect. The UK Statistics Authority, the Government Analysis Function, the ICO and HM Treasury publish nothing of the kind. UK reference material on campaign and media measurement comes instead from industry bodies such as IAB UK, the IPA and ISBA, and from supplier and agency reports. None of those sources is official statistics, so each enters your report only through the test below.

A global supplier rate or broader UK campaign statistic cannot fill that gap. The appropriate output is an admissibility record and, where comparison fails, "no usable benchmark".

What to take away

Describe what the external figure measures

Open the original publication rather than a summary. Record publisher, funding, research date, geography, buyer and creator population, selection frame, sample construction. Record platform and service configuration, product and campaign category, channel, content format, paid or earned state, observation window and correction date. Mark whether the record is official evidence, independent research, supplier analysis or a supplier claim.

A figure offered by a platform such as Modash, HypeAuditor or CreatorIQ usually comes from that supplier's own creator database, from a survey of its own members or clients, or from both, and the method note or data appendix for it may sit in a separate document from the headline report. Ask which source it is. Platform-observed events describe the supplier's population, a member survey describes its respondents, and neither describes the open market; where the report does not say, record the field as not disclosed.

The Government Analysis Function advises producers to explain quality and uncertainty, including coverage, completeness and limitations. Its audience is official-statistics producers, but the disclosure fields suit a commercial evidence review.

Official-statistics governance is not the only test. The MRS Code of Conduct and the ICC/ESOMAR International Code on Market, Opinion and Social Research set the professional standards for commercial research in the UK, and both expect a producer to describe the purpose, method, sample and limitations of a study. A creator-platform benchmark that names a rate but not its sample frame, or a sample but not its fieldwork dates, fails that test before any comparison begins — and the same expectation applies to the supplier report, the agency analysis and the published survey alike.

Rebuild the denominator

Copy the external numerator and denominator definitions exactly. Identify whether the unit is a profile, account, person, represented creator, proposed contact, message, response, contract, asset, publication, view, order or customer. Find the identity and deduplication rule. Record timezone, period, exclusions, missing and late-data treatment, weights, imputation, confidence or other uncertainty, and whether later corrections changed the release.

A profile found cannot be compared with an admitted record. Contacted creators differ from messages sent, and silence cannot be labelled unwillingness. Paid placement differs from an earned editorial response. Platform-attributed activity is neither reconciled retained value nor causal effect.

The UK Statistics Authority's current official-statistics standards require suitable data and transparent quality. They do not convert a private platform report into official statistics, but they provide a useful challenge: is the source fit for the intended comparison?

Examine visibility and permissions

State which events the platform could observe and which occurred outside it. Document API or contract limits, data loss, account deletion, blocked content and later corrections. Record the privacy authority for the buyer's own use of personal data. Do not add inferred age, residence, protected traits, interests, authenticity, suitability, representation, audience composition, reach or willingness to force closer segmentation.

The ICO's Direct marketing guidance says public social-media information is not automatically free for direct-marketing use. The UK GDPR and the Data Protection Act 2018 govern relevant personal-data processing. A published benchmark does not supply authority for outreach or remove safeguarding, rights, advertising, security and accessibility gates.

Apply a compatibility decision

Create one row for each field above and mark compatible, materially different, unknown or not disclosed. Name the reviewer and record the reason. If population, identity, denominator, geography, period or campaign state is materially different, do not pool or directly compare the rate. If visibility, permission or missingness is unknown, hold the comparison pending evidence.

The record below is the form to complete for the benchmark you are judging: carry one row for each figure you want to admit, fill the fields from the report itself, and name the reviewer. The two rows show the shape of a compatible decision and an incompatible one.

The 2025 AQuA Book advises that analytical assumptions have an owner and lifespan and that uncertainty is inherent. Use that discipline to set an evidence expiry and recheck trigger. Do not borrow its government status as approval of this private method.

Keep an external rate away from the target field

An admissible external observation may provide context or a range for sensitivity analysis. It still does not set the buyer's service threshold, campaign expectation or forecast. Leave the threshold [ ] for the buyer to approve from its decision, risk, baseline and current evidence.

Causal comparability requires more than matching labels. HM Treasury's Magenta Book distinguishes process monitoring from impact evaluation. Without a credible counterfactual and compatible design, describe association or attributed activity only. Preserve the rejection record so a later report cannot quietly reintroduce the unsuitable benchmark.

Before you act

Common questions

What should I record when describing an external figure?

Record the disclosure fields that establish provenance: publisher, funding, date, geography, populations, sample construction, platform configuration, category, channel, format and evidence type. Check whether those fields match the target report before any comparison.

Why can a profile found not be compared with an admitted record?

The unit and denominator definitions do not match. Record the identity, deduplication and observation rules before any comparison.

When should I hold a comparison pending evidence?

Hold it when visibility, permission or missingness is unknown. If population, identity, denominator, geography, period or campaign state differs, record an incompatible decision instead. Name the reviewer, record the reason, and set an evidence expiry and recheck trigger.