Editorial draft available to read. Factual and language review is pending. Examples are illustrative.
Define a change unit
Record dates, pages, owners, rationale and versions for meaningful content, structure, configuration and measurement changes. Not every typo needs equal detail; define relevance to offer, access, intent or experience.
Preserve context and verification
Add the prior hypothesis, implementation evidence and verification issues, alongside campaigns and business changes. Do not rewrite hypotheses to fit outcomes. Separate expectations before the change from learning afterwards.
Use the log in review
Example: decline coincides with route changes and measurement outage. Investigate both before blaming content. The log should expose changes and deployment evidence with alternative hypotheses; timing guides investigation but does not prove causation.
Key takeaways
- Record what changed and when to interpret observations without inventing causality.