Your blocklist is a business decision you can't see

Brand-safety keyword blocklists silently decide which journalism gets ad funding. This scorecard audits a blocklist against a small sample of legitimate, openly-licensed news — health, conflict, LGBTQ+ coverage, finance — and shows exactly what it would defund, with honest small-sample uncertainty (95% Wilson intervals).

Keyword block lists flag 63% of Newsweek's inventory; advertisers blocked over 40% of The Washington Post's inventory on brand-safety grounds (The Current, Feb 2025). Brand-safety providers disagree on the same article in 22.98% of two-provider cases and 41.57% of three-provider cases; researchers argue "a unified and transparent framework is needed" (Smith et al., arXiv:2601.01303). The four-category breakdown below is this audit's design, not a claim from either source.

Illustrative blocklist scorecards

Each list is scored against the same corpus: real headlines + ledes (public-domain / CC-licensed; sources in the footer). A "block" means the list's keywords match the headline or lede — the same blunt mechanism used in practice.

Audit your own blocklist

Paste one keyword or phrase per line (# comments ignored).

No data leaves your browser. This page is a single static file with no server calls, no analytics, and no network requests — scoring runs entirely client-side. Reload the page and your list is gone.