29-Node GSC Drop Triage & 4.4M URL Index Governance
Automated Search Console anomaly detection across 5 dimensions, 4.4M sitemap index-bloat diagnostics, and a 17-check helpful-content LLM audit running for under $1/day.
STAR Engineering Breakdown
01 · Situation & Problem
Organic traffic drops were discovered late and diagnosed manually, while 4.4M sitemap URLs suffered severe index bloat (86% of job URLs earned zero impressions while job pages drove 69% of clicks).
02 · Technical Execution
Built a 29-node n8n triage workflow firing on >10% click or >15% impression drops, decomposing losses across page, query, country, device, and search appearance before URL-inspecting the top 5 losers.
03 · Index & Cloud Governance
Traced slow job indexing to cookie-gated caching, specified pruning 2M+ dead-weight URLs, reconciled GSC row-limit truncation with a week-over-week baseline, and cut marketing GCP spend ~98% ($7.5K/yr) by locking down runaway Maps API calls.
04 · Verified Impact
Continuous daily anomaly monitoring ran from Dec 2025 to Aug 2026 alongside a 17-check automated helpful-content audit that graded pages for under $1/day and auto-filed monday.com engineering/editorial tasks.
def should_trigger_triage(curr: GscMetrics, base: GscMetrics) -> bool:
click_delta = (curr.clicks - base.clicks) / max(base.clicks, 1)
impr_delta = (curr.impr - base.impr) / max(base.impr, 1)
is_anomaly = click_delta <= -0.10 or impr_delta <= -0.15
return is_anomaly and curr.rows_complete # Guards row truncationBenchmark Table
| METRIC | MANUAL BASELINE | PROGRAMMATIC ENGINE |
|---|---|---|
| Traffic Drop Root-Cause Triage | 4+ Hours Manual CSV Diffs | Automated 29-Node Slack Report |
| Crawl Budget & Sitemap Hygiene | 4.4M URLs (86% 0-Impression) | 2M+ Prune Spec & Cache Fix |
| Marketing Google Cloud Spend | ~$7,650/yr Uncapped API Usage | -98% Reduction (~$7.5K/yr Saved) |