Measurement
Part of Fitness creator measurement in England: defining health metrics from the CMO activity guideline
Three fitness creators measurement mistakes that overstate health evidence
Three fitness creators measurement mistakes overstate health evidence: testimonials as proof, before-and-after as cause, and health data kept for colour.
What to take away
- A favourable comment is neither a validated measure nor a comparison group, so it cannot carry an outcome claim.
- Before-and-after material cannot show that an intervention caused the difference.
- Fitness-tracker information can reveal health status, so keep participant records out of a general marketing view.
- A health reviewer, an evaluator and a privacy owner each hold a gate that a report must pass.
- If a report blurs these boundaries, hold the report.
Where the three mistakes sit
Three measurement mistakes turn weak material into health evidence: a testimonial read as proof, a before-and-after pair read as cause, and tracker data collected only to add colour. Each has a different owner who can stop it, and a different repair.
| Mistake | Why it misleads | Who stops it | Repair |
|---|---|---|---|
| Testimonial as outcome evidence | Not a validated measure or a comparison group | Health reviewer | Keep comments in the audience panel and outcomes in a reviewed protocol |
| Before-and-after as attribution | Selection, time, concurrent activity and measurement can all explain a difference | Evaluator | Predefine population, comparison, outcome, time points, analysis, uncertainty and stop conditions |
| Health data kept for colour | Tracker information can reveal health status | Privacy owner | Document necessity, legal conditions, access, retention and disclosure control before collection |
Mistake 1: turning testimonials into outcome evidence
A favourable comment is neither a validated measure nor a comparison group. The Committee of Advertising Practice (CAP) says in its advice on claims in testimonials and endorsements that testimonials alone do not constitute substantiation, so marketers should not rely on them as support for claims.
Keep comment moderation in the audience panel and keep health outcomes in a reviewed protocol. A health reviewer stops publication when a testimonial is being used to imply efficacy. The same rule applies to a report: a quote placed beside a chart still implies the chart proves the quote.
When a report needs an outcome, ask what measure was defined before the comment arrived. If none was, the honest label is feedback. The return on investment article explains why health outcomes stay out of the numerator.
Mistake 2: reading before-and-after material as attribution
Two observations can differ because of selection, time, concurrent activity, measurement or many other factors. The Magenta Book, the government's guidance on designing evaluations, says monitoring change does not establish that an intervention caused it.
The evaluator must predefine the population, the comparison, the outcome, the time points, the analysis, the uncertainty and the stop conditions before any observation is taken. Without those, a pair of images or a before-and-after score is a story, not a result.
Images need separate permission and advertising review, and they do not repair the design. A tagged sale is a different question, since attribution methods describe where a visit came from, not whether an exercise worked.
Mistake 3: collecting health data for dashboard colour
Fitness-tracker information can reveal health status. ICO special-category guidance lists data from fitness trackers among examples of health data. Do not import participant records into a general marketing view just to make a panel look richer.
The privacy owner documents necessity, legal conditions, access, retention and disclosure control before collection starts. A change of purpose, or a small-cell risk where a group is so small that a person could be recognised, triggers review and possible deletion. The data protection guide covers the wider rules for sign-up, screening and marketing.
One question before a health sentence goes out
For every headline number, ask whether another analyst can reconstruct it and whether the sentence makes a stronger claim than the design. Operations data can reveal a failed gate. Audience data can describe behaviour. Attributed sales can follow a declared rule.
A causal health statement needs a suitable evaluation and qualified interpretation. If those boundaries are blurred, hold the report. A dashboard that keeps health panels apart makes the boundary visible, and the quality checklist shows the gates a release must pass.
- Label comments as feedback unless a measure was defined first.
- Keep testimonials out of any outcome or efficacy statement.
- Predefine comparison, outcome, time points and stop conditions for any before-and-after work.
- Document necessity, legal conditions, access and retention before collecting health data.
- Name the health reviewer, evaluator and privacy owner for the report.
Common questions
Can a testimonial support a health claim?
No. CAP advice says testimonials alone do not constitute substantiation. A health reviewer stops publication when a testimonial is used to imply efficacy, and objective claims need evidence of their own.
Why can before-and-after images not prove a result?
Two observations can differ because of selection, time, concurrent activity or measurement. Monitoring change does not establish that an intervention caused it, so the design must be set before anyone observes anything.
Is fitness-tracker data health data?
It can be. Fitness-tracker information can reveal health status, so treat it as potential health data. The privacy owner records necessity, legal conditions, access and retention before collection begins.
Who owns each gate?
The health reviewer owns testimonial claims, the evaluator owns before-and-after design, and the privacy owner owns health data collection. The team roles list shows who can hold each gate; name each person before the report is drafted.
