Methodology & Sourcing
Dated, source-labeled records are the credibility standard here. This page explains which fields readers should expect and how missing evidence is handled.
Sourcing standards
A record is treated as campaign-grade only when it includes a campaign type, named platform, spend range, primary metric, measured value, and review context. Broader industry figures remain labeled as context rather than being presented as operator campaign performance.
Example of that labeling in practice: Do AI Job-Loss Warnings Actually Apply to Advertisers? shows our own metric next to any cited industry figure, kept visually and textually distinct.
Last-reviewed policy
AI ad defaults change quickly and often without much warning. Benchmarks, Bidding, and Creative records display last_reviewed_at when the editorial record supplies it. A missing or old date is a reason to treat the figure as potentially stale and check the Tracker for relevant changes.
Corrections & disputes
Every detail page across Benchmarks, Tracker, Bidding, and Creative carries a correction module after the full body. Submissions there are meant to report a corroborating or contradicting result from your own account, or to flag a factual inaccuracy — not to open a public discussion thread. If you need a direct reply, use Request an Audit.