Three points to take away
- 01
Link every figure to its original table or passage.
- 02
Separate platform, third-party, and case-study evidence.
- 03
Do not transfer averages directly to your account.
Claim-evidence table
Record claim, source, publication date, sample, market, metric definition, limitations, and incentives. Mark conclusions as pending when key fields are missing.
Translate into a local test
Design a bounded incremental test with your audience, creative, and attribution window, then use local results rather than repeating an average uplift.
Verification checklist before action
Open every original source and confirm the summary preserves qualifiers and metric definitions.
Run a bounded trial with non-sensitive samples and retain successes, failures, and human corrections.
Before wider use, name an owner, data boundary, stop condition, and review date.