A fill-in monitoring + optimization workbook for the deployed ScamCheck growth engine: what to track for indexing, Discover, SEO/GEO, content, distribution, monetization, and cost — where to read it, healthy thresholds, and a weekly optimization loop seeded with computed publish/backlink/conversion priorities. Guardrails keep it low-cost and stable.
Post-launch operating workbook. Live metrics (indexing counts, Discover impressions, RPM) live in your GSC / AdSense / Vercel / Analytics — this doc tells you exactly what to read, where, the healthy range, and what to do. Sections seeded with priorities computed deterministically from the engine (no live data needed). No new systems — monitoring + high-ROI optimization only.
Note: the engine's
/api/healthreports live AI/persistence mode. In production it should readai: live,persistence: firestore,env.ok: true. (A local sandbox reads mock/memory — that is expected and not the production state.)
Publish priority — topic leverage (demand × RPM × momentum):
| Track | Healthy | Action if not |
|---|---|---|
| Indexed page growth | rising toward ~250 over 2–3 wks | request-index stalled tier-1 |
| Excluded / "Discovered – not indexed" | shrinking | improve internal links / content depth |
| Sitemap coverage | submitted, ~730 discovered | resubmit; check origin matches canonical |
| Crawl anomalies (5xx/redirects) | none | check Vercel logs / canonical config |
| Orphan risk | 0 | re-run crawl-depth after big content adds |
| Track | Healthy | Action |
|---|---|---|
| Discover impressions | first by wk 2–3, then rising | push freshest tier-1 + in-season hub |
| Discover CTR | improving | strengthen headline (honest) + OG image |
| Strongest Discover pages | identify top 5 | make more like them; interlink |
| Freshness decay | refresh pages older than ~30d | update "last updated" + trending strip |
| Image/OG effectiveness | large-image previews showing | confirm robots allows /api/og |
/api/scam-intel/growth?view=entities (fastest-growing categories/platforms/regions).| Track | Where | Use |
|---|---|---|
| Shorts / LinkedIn / X engagement | each platform's native analytics | double down on top hooks (from alert-formats) |
| Backlink growth / referring domains | GSC → Links + GA/Plausible referrals | scale the winning outreach angle |
| Widget embeds | referral traffic from widget.js loads | pitch more sites in that vertical |
| Track | Healthy | Action |
|---|---|---|
| Vertex spend today | under DAILY_BUDGET_USD | breaker auto-halts; lower caps if near |
| Quota utilization | under 85% | back off autopilot frequency |
| Firestore reads/writes | low (counters, cached snapshot) | investigate spikes; avoid scans |
| Cache hit rate | rising | ensure cache TTL not bypassed |
| Queue efficiency / DLQ | failure under 5%, DLQ 0 | inspect + replay from DLQ |
| Unnecessary AI usage | none (dedup + quality gate) | confirm crons early-exit when idle |
Run weekly; most inputs are one API call or one GSC view.
/api/scam-intel/growth?view=leverage → publish/promote top movers (baseline: Investment, UPI, KYC, Phishing, Fake-job).?view=backlink (baseline: UPI, OTP, KYC) → pitch the widget to the matching vertical.DAILY_BUDGET_USD, AUTOPILOT_PER_RUN), not bigger infra.Share a live origin URL (I'll run the post-deploy smoke + crawl/canonical/OG/schema checks against production) and/or exported GSC/AdSense numbers (indexed count, top queries/pages, Discover impressions, RPM) — then I'll produce a data-driven weekly report: weakest pages, CTR-hook targets, refresh list, and the next week's priorities.