Score any headline for Google Discover

The score blends two signals: 60% empirical (the average traffic of the 20 articles most semantically similar to your title in our 500,000-article Discover corpus) and 40% editorial heuristic (length, opening, emotion verb, recent year mention, etc.).

To deliver a deep analysis: semantic embedding, pgvector search, percentile computation and real examples — each analysis takes several seconds of compute. Better than a sloppy multi-title comparator.

It's continuously indexed from Google Discover (French and English feeds), with their real traffic metrics. Embedding vectors are stored in Postgres + pgvector for semantic search in milliseconds.

It's a strong indicator, but Discover also factors in your site's authority, freshness, and technical quality (Web Profile, image, schema). Our Profiler complements the analysis on the technical side.

From the 20 similar articles we extract the median and P90 of their real Discover views, then apply a ×500 coefficient to estimate the traffic projectable to your site. It's an order of magnitude, not a promise.

Yes — 3 free analyses per day, no sign-up. For larger volumes or API access, the Pro plan starts on quote.