Measurement
Part of Influencer agencies measurement: populations and an evidence ladder
How to vet an influencer-agency benchmark figure in England
How to vet influencer agencies benchmark research in England: what ASA, platform and ONS sources cannot support, nine fields to request, a release rule.
No reliable, directly comparable England benchmark for influencer-agency response, publication, engagement, conversion or return was identified in the authoritative source universe reviewed on 6 September 2026. That is the starting result, not a blank to fill with a supplier average.
Until a proper study exists, an England buyer can use its own frozen baseline, but must not label it a market norm. The sources that do exist cover recognition and definitions, which is a different question from agency performance.
What to take away
- No reliable England benchmark for influencer-agency performance was found in the reviewed sources.
- Existing sources cover ad recognition or platform definitions, not agency campaign delivery or sales.
- A convenience average of willing contributors carries errors that a narrow interval does not cure.
- Release a figure only when its fields are comparable enough for the stated decision.
- If comparability fails, return no usable benchmark and document the failed fields.
What each reviewed source covers
Each source answers a narrower question than a benchmark does. None samples England agency campaigns, and none gives a response, publication, conversion or return rate. The table sets out what each one supports and where it stops.
| Source | What it covers | What it cannot support |
|---|---|---|
| ASA's February 2026 consumer research on influencer marketing | UK advertising recognition and disclosure, primarily on Instagram and TikTok | A response, conversion or return benchmark for England agency campaigns |
| YouTube's global audience help | Unique viewers, defined as an estimated field | A common England denominator across creators, platforms or advertisers |
| Google's explanation of the conversion window | Which post-interaction events are recorded | A common denominator across advertisers |
| ONS quality definition | Coherence and comparability, as a review lens | An official statistic; this benchmark is not one |
| Supplier records | The supplier's own definitions | A like-for-like comparison with other records |
Whether participants recognise an advert cannot stand in for delivery or sales performance. The advertising rules for influencer agencies cover recognition, claims and placement, which is where the ASA material belongs.
Why an average of willing contributors misleads
A sample figure is not the population value. ONS explains that the difference between a sample statistic and the population value is caused by sampling error and non-sampling errors. A convenience panel of willing agencies may carry selection and reporting error that a narrow confidence interval does not cure.
ONS says coherence and comparability concern similarity across sources, methods, time and geographic domains in its quality definition. Use those dimensions as a review lens, while making clear that a proposed benchmark is not an official statistic.
Do not compare engagement ratios until each source uses the same eligible event and denominator. Keep platform-reported estimates separate from directly observed campaign records. Published ratios by platform are covered in engagement rate benchmarks.
Commercial comparisons need one accepted outcome, cancellation cut-off, contribution definition and full-cost boundary. Causal impact requires a credible counterfactual rather than an average of attributed results.
Nine fields to ask any figure's owner for
Anyone who offers a benchmark should be able to show these fields for each measure. If they cannot, the figure is a claim, not a benchmark.
Measure specification checklist
- Event and eligible population
- Numerator, denominator or unit
- Reporting dates and timezone
- Creator, audience, platform strata
- Source system and query version
- Duplicate, missing, late treatment
- Privacy and suppression rules
- Sampling or other uncertainty
- event and eligible population;
- numerator, denominator or unit;
- reporting dates and Europe/London timezone;
- creator, audience, platform and format strata;
- source system and query version;
- duplicate, missing, late and corrected-data treatment;
- privacy and suppression rules;
- sampling or other uncertainty;
- data owner and decision owner.
The buyer then adds its own threshold and update trigger. The key metrics for influencer agencies show how fixed events and denominators make that possible.
When a figure can be released
Release a benchmark only if population, geography, platform, format, period, denominator and data rights are comparable enough for the stated decision. Otherwise return "no usable benchmark", document the failed fields and set a recheck date.
This evidence-gap conclusion is more useful than a precise-looking number assembled from incompatible UK, global and supplier records. Spend totals have the same problem, as the note on why England has no market total explains.
- Name the specific decision the figure informs.
- Confirm the population and sampling frame come from permissioned records.
- Check that eligibility and exclusions were set before outcomes were seen.
- Obtain the measure, denominator and uncertainty method.
- Have a statistician and privacy reviewer approve the design and release table.
- Apply the release rule and document any failed fields.
The full set-up, from events to reporting, sits in the measurement and reporting guide.
Common questions
Why is there no reliable England benchmark for influencer-agency performance?
The reviewed sources measure other things: ad recognition, platform definitions or a supplier's own records. None samples England agency campaigns, so none gives a comparable response, publication, engagement, conversion or return figure.
Can a buyer use its own baseline instead?
Yes, a frozen baseline of its own campaigns. It must not be labelled a market norm.
What does the ASA research tell a buyer?
It studied UK advertising recognition and disclosure, primarily on Instagram and TikTok. That is relevant to a recognition study, but it does not give a response-rate, conversion or return benchmark.
When should a benchmark be released?
When its fields match the stated decision closely enough. If they do not, the honest answer is no usable benchmark, with the failed fields written down and a recheck date set.
