in/metehanyesilyurt | metehan.ai | @metehan777
Powered by Claude Opus 4.6 + SDK Telemetry

Discover pCTR Predictor

Predicts Google Discover click-through rates using a model calibrated against actual SDK telemetry — entity matching, freshness buckets, and the blindness penalty (β).

📐 Discover pCTR Formula (SDK-Derived)

quality = Σ(wᵢ × fᵢ) 8 quality dimensions β_penalty = 1 − 0.35 × (clickbait_score / 10) raw = quality × β_penalty pCTR = 0.5% + (22% − 0.5%) × σ(0.65 × (raw − 5.5))
β Blindness Factor: From Google's LTV formula (LTV = bid × pCTR − β). Clickbait titles trigger high initial CTR but low engagement_time_msec. The feedback loop (Stage 9) reduces future pCTR. Our β_penalty models this: a clickbait_score of 7/10 reduces quality by 24.5%.
Entity Density22%
Topic Clarity18%
Informational Value16%
Freshness Signal12%
Engagement Depth10%
Title Formatting8%
Natural Authority8%
Visual Promise6%
Multilingual — works with any language. Try titles in English, Turkish, German, Spanish, or any other language.
📊

Enter news titles to predict their Google Discover click-through rate.