Q-09 Publish first-party data and explain where it comes from

Why it matters

Data shown without its sample, period or scope cannot be judged, so readers have no way to tell whether it applies to them. Search engines and AI answer engines also have little reason to trust or cite figures that arrive without context. Publishing raw data before data protection duties are settled creates a separate risk.

How to implement

1. List the anonymous operational, sales, service, usage, survey or test data you could use.
2. Resolve personal data and data protection obligations before publishing anything raw.
3. State the sample size, period and inclusion and exclusion criteria next to each dataset.
4. Present aggregate results separately from individual cases.
5. Add a version and date system for any data that is updated.

Checklist

• Inventory the anonymous operational data that can be used.
• Do not publish raw data until personal data and data protection obligations are resolved.
• Write down the sample size, period and inclusion and exclusion criteria.
• Distinguish aggregate results from individual cases.
• If the data is updated, set up a version and date system.

How to audit

• Check that a data provenance record exists.
• Check the sample size and methodology.
• Review the data for sensitive information.

Related controls

• Q-04 Content must be non-commodity and hard to reproduce P0
• Q-10 Research and benchmark methodology must be understandable P1
• Q-11 First-hand experience claims must be backed by evidence P1
• Q-12 Comparisons must use shared criteria and real trade-offs P1
• Q-13 Calculators and selectors must deliver real utility P1

Sources

• [G3] Optimizing for generative AI features on Google Search
• [B1] Bing Webmaster Tools: AI Performance