Measurement
Part of Fitness creator measurement in England: defining health metrics from the CMO activity guideline
Four fitness creator metrics, each with its denominator and its limits
Define fitness creator metrics by evidence lane, with complete denominators, data sources, uncertainty, owners and clear health-outcome limits.
What to take away
- The four metrics are review-gate completion, eligible thumbnail click-through rate, validated campaign order rate and health outcome change.
- Each one needs a stated population, numerator, denominator and unit before collection starts.
- No benchmark figures appear here, by design.
- Health outcome change needs a pre-approved protocol, and health information can be special category data.
The four metrics at a glance
Four records cover the same ground: what was counted, out of what, in which unit and over which period. No benchmark is given here, by design. Fitness creator benchmark research shows how to build a baseline first.
| Metric | Numerator over denominator | Unit |
|---|---|---|
| Review-gate completion | Scheduled items published with every signed review, over all scheduled items | Percentage, Europe/London |
| Eligible thumbnail click-through rate | Eligible impressions followed by a view, over all eligible registered impressions | Percentage |
| Validated campaign order rate | Eligible settled orders, over consented sessions carrying the campaign identifier | Orders per 100 sessions |
| Health outcome change | Result on the validated measure, over eligible participants under the protocol | Score or rate with interval |
Review-gate completion
The population is every item scheduled for publication in the Monday-to-Sunday reporting period. The numerator counts scheduled items whose published version carries every required signed review.
The denominator is all scheduled items, including late or withdrawn releases, but not tests cancelled before scheduling approval. Report a percentage in Europe/London time from versioned records.
The operations lead owns the measure. Incorrect risk tags or mismatched approvals create uncertainty. A missing clinical or advertising gate triggers a same-day hold and record review.
Eligible thumbnail click-through rate
For one YouTube format and surface, the population is registered thumbnail impressions in a stated month. The numerator is impressions followed by a view under YouTube's definition. The denominator is all eligible registered impressions, in per cent. YouTube's official metric definition is the current source.
Include only the named videos, the surface and the country filter. Apply the account timezone, which may differ from Europe/London. A movement prompts a surface check, not a health claim.
Validated campaign order rate
The population is unique, consented landing sessions carrying the agreed campaign identifier during a declared period. The numerator is eligible settled orders matched to those sessions inside the stated window. The denominator is those sessions, reported as orders per 100 sessions. Use Europe/London order time.
Exclude staff, tests, cancellations, refunds, duplicates and unmatched orders. Google's campaign URL guidance documents how UTM values populate acquisition reporting, but a tag does not prove causation. Call the result attributed activity, not incremental sales.
The commercial analyst owns the metric and revises it after a refund reconciliation. Missing consent, shared devices and unmatched orders add uncertainty. Separating real effect from coincidence needs a comparison design, and attribution methods that prove what actually worked sets it out.
Health outcome change
Do not put a generic version on a marketing dashboard. Population, numerator or score, denominator and time points come from a pre-approved protocol. Timezone, England eligibility and exclusions come from that protocol too.
The evaluation lead owns the analysis, and a registered health professional owns the clinical interpretation. Report missing data, selection, attrition, measurement error and confounding beside the estimate and its interval. The Magenta Book explains why movement alone does not establish cause.
Health information can be special category data. The ICO treats fitness-tracker information as data concerning health. The privacy reviewer must approve purpose, legal conditions, access and retention before collection starts.
Example: working the thumbnail arithmetic
Suppose a channel registers 40,000 eligible impressions in March across the named videos and surfaces. YouTube records 2,000 views from those impressions under its definition. The arithmetic is 2,000 divided by 40,000, which is 0.05, or 5 per cent.
That 5 per cent describes one surface in one month. Publish the month, timezone and filter set beside the figure. It is not a target and says nothing about content quality.
Where disclosure and claim rules bite
These metrics govern what you count, while the advertising rules govern what you may say. The CAP Code and the Consumer Protection from Unfair Trading Regulations 2008 catch misleading claims and fake endorsements. The NHS content standard supports documented approval for health content. ](/fitness-creators-advertising-rules-in-england) works through the disclosure and health-claim tests for England.
Before you act
- Fix the population, numerator, denominator and unit before collecting data.
- Disclose the timezone and name the source system for each metric.
- Get privacy reviewer approval before collecting identifiable health data.
- Treat a missing approval or a safety signal as a stop trigger.
Common questions
Which metric needs a registered health professional?
Only health outcome change. The evaluation lead owns the analysis, but clinical interpretation sits with a registered health professional.
Can a thumbnail click-through rate fall with nothing wrong?
Yes. Viewer mix, surface intent and seasonality all move it. Treat a fall as a prompt to check the filter set and the surface, not as a verdict on the content.
What changes the campaign order rate denominator?
Consent status and the matching window. Widen the window or count unconsented sessions and the population changes.
