Import the existing local project snapshot. This commit organizes existing files; it does not represent newly developed functionality.
3.9 KiB
SEO Recovery Plan — June 2026
Status: ACTIVE — started 2026-06-10 Owner: آرسام صباغ · Tracking doc for the organic-traffic recovery after the June 3 demotion
Diagnosis (confirmed 2026-06-10)
On June 3, 2026 we published 36 articles (art38–art73) in a single deploy, doubling the site overnight, with backdated publish dates. Google's crawl spiked to 166 req/day, the site was re-evaluated, and a site-wide algorithmic quality demotion followed:
ai-data-centric-2026— formerly ranked #1 above Wikipedia — has had zero Google referrals since June 5 (10 days of logs).- Real organic clicks dropped to ~3–5/day (homepage brand queries only).
/profile/@radman-ghelichiwas the #2 organic landing page; the author purge on June 8–10 deleted it (now 301 → /authors).
Verified NOT the cause: canonicals, noindex, sitemap health, crawl errors — all clean. Googlebot crawl is steady at 80–100/day → no manual action. This is algorithmic and recoverable.
Data caveat: ~90% of nginx "views" are scanners/bots (wp-admin probes, spoofed google.com referrers). The metric that matters is Search Console impressions/clicks, never raw nginx hits.
Phase 0 — Stop the bleeding (June 10–14)
| # | Item | Owner | Status |
|---|---|---|---|
| 0.1 | Churn freeze: after the single consolidated deploy below, no URL changes, no author surgery, no bulk publishes, no sitemap regeneration for 2–3 weeks | both | ☐ in effect after deploy |
| 0.2 | Google Search Console audit: check Manual Actions (expect none), Performance → compare 7d vs prior 28d, find exactly which queries/pages lost impressions and the exact drop date | USER | ☐ pending |
| 0.3 | No backdating, ever: every future article carries its true publish date | both | ☐ policy |
Phase 1 — Rebuild trust signals (June 10–24) — one consolidated deploy
| # | Item | Status |
|---|---|---|
| 1.1 | Author E-E-A-T: real bios + sameAs links (asabbagh.com, arsamsabbagh.ir, verified socials) on author pages and in Person JSON-LD for the 2 remaining authors |
☐ implementing now |
| 1.2 | Thin-content repair: expand the weakest old articles (art1–art15 range, 600–900 words) to the proven formula — <p class="lead">, numbered H2 sections, <strong> entities, bilingual terms, 8+ sections / ~2500+ words, factCheckedAt. Content improvement of existing URLs is a recognized recovery tactic and adds no new URLs |
☐ implementing now |
| 1.3 | Internal linking: ensure related-article links are real crawlable <a href> in the static article HTML (not just client-side React), so the 73 articles stop being islands |
☐ implementing now |
| 1.4 | Run npm run check:seo → full npm run build → reload nginx — one deploy containing 1.1–1.3, then freeze (0.1) |
☐ |
Phase 2 — Earn signals back (June 24 →)
| # | Item | Owner |
|---|---|---|
| 2.1 | Cadence, not bulk: 2–3 new articles per week, real dates, never batch-publish again | USER + Claude |
| 2.2 | Distribution: share each article on Virgool, Persian tech Telegram channels, LinkedIn. Target: 2–3 backlinks from Persian tech sites within a month | USER |
| 2.3 | Track recovery weekly in GSC: impressions trend, ai-data-centric-2026 ranking, new-article indexing speed |
USER |
| 2.4 | After freeze lifts: re-evaluate remaining thin articles — expand or prune | both |
Expectations
Algorithmic demotions lift weeks to months after the site stabilizes. The domain has already proven it can rank #1. The single most dangerous move during recovery is another bulk change.
Recovery checkpoints
- Week 1 (Jun 17): GSC shows no further impression decline
- Week 3 (Jul 1): new weekly articles indexed within 48h
- Week 6 (Jul 22):
ai-data-centric-2026reappears for its core query; impressions trend ↑ - Month 3 (Sep): organic clicks exceed pre-drop baseline