content-refresher

# Content Refresher ## Decay score formula ``` decay = 0.30 × traffic_drop_pct + 0.25 × rank_drop_normalized + 0.15 × ctr_decrease_pct + 0.15 × freshness_age_normalized + 0.15 × replacement_pressure ``` ## Data sources (real executors — Rule 6) 3 of the 5 inputs come from the connected first-party Google data — pull them per project (IDs live in `projects/{slug}/business-context.json :: analytics`): ```bash # traffic_drop_pct ← GA4 organic sessions over the window python -m scripts.audit.ga4_fetch --property-id {analytics.ga4_property_id} --days 28 --json # rank_drop_normalized + ctr_decrease_pct ← GSC position/CTR per URL over time python -m scripts.monitor.gsc_api_ingest --site {slug} --mode last-28 --json ``` `freshness_age_normalized` = days since dateModified; `replacement_pressure` = SERP churn for the target keyword (`scripts.fetch.serpapi_query --engine google`). Degrade gracefully: a project whose GA4 property is brand-new (0 traffic rows) scores `traffic_drop=0` until data accumulates — empty GA4 is NOT a crash. Both Google sources authenticate through `scripts/_core/google_creds.py` (the unified token), so no per-call credential setup is needed. | Score | Action | |---|---| | <30 | URGENT — refresh within 7d | | 30-50 | Refresh in 7d | | 50-75 | Refresh in 30d | | >75 | Healthy; defer | ## Action routing (don't always full-rewrite) Before doing a full rewrite, run the decision router — it diagnoses the signal and names the MINIMAL su...