clownnvd/seo-geo-on-top

SEO + GEO: On Top

Get content to page 1 of Google AND get quoted by AI answer engines (Google AI Overviews, ChatGPT, Perplexity, Gemini, Bing Copilot). Evidence over theory — this skill leads with what actually ranked SpeakShark, and stays true via a living field log you update after every real result.

Prime directive

1. Proven > plausible. A tactic with a dated win in references/field-log.md outranks any "best practice". Treat unlogged theory (including other geo-* skills and this skill's own theory sections) as a hypothesis to test, not a law.
2. Verify live, never assume. Check the real rendered page with curl/GSC, not a summarizer — small models miss <script ld+json> and report false negatives (this happened during the SpeakShark audit: schema was reported "missing" when 8 JSON-LD blocks were live).
3. One variable at a time when testing causality, so the field log stays honest.
4. Log every result. The log is the asset; this playbook is just its summary.

How to use this skill (full reference map)

• Picking what to target (keyword + intent + winnability) → references/keyword-research.md.
• Writing/optimizing a post → references/content-onpage.md + references/geo-ai-engines.md, then the checklist below.
• Shipping/fixing schema → references/schema-cookbook.md (paste-ready JSON-LD).
• Indexing / rendering / hreflang / CWV / crawlers → references/technical-seo.md.
• Multilingual (19 locales) + local → references/local-international.md.
• Getting cited by AI engines → references/geo-ai-engines.md.
• Brand not recognized by AI / empty sameAs → references/entity-sameas.md.
• Links, digital PR, brand mentions, authority → references/offpage-authority.md.
• YouTube/Reddit/social/repurposing + SERP features → references/distribution-multichannel.md.
• Surviving Google updates, avoiding penalties, recovery, AI-content rules → references/updates-penalties-recovery.md.
• Converting the traffic + analytics/attribution → references/cro-analytics.md.
• Tracking results + updating the playbook → references/measurement-iteration.md.
• Evidence of what worked → references/field-log.md.
• The exact live, working stack + file map → references/proven-stack.md.

Covers every aspect: keyword research · on-page · technical · international/local · schema/structured data · GEO/AI-engine citation · entity/sameAs · off-page/links/PR · multichannel distribution + SERP features · algorithm updates/penalties/recovery · AI-content safety · CRO · analytics/measurement.

Proven levers (verified — field-log 2026-06-06)

Ranked #2 on Google + quoted verbatim by Google AI Overview for the high-intent query "best english speaking apps 2026". By confidence:

| Lever | Why it worked | Conf. | |-------|---------------|-------| | Freshness + year in title | "best X 2026" triggers query-deserves-freshness; publish + index inside the window | HIGH | | Exact-intent listicle + comparison table | matches commercial-investigation intent; tables are AI-extractable | HIGH | | Credible E-E-A-T framing | "tested across…", honest "no app wins all", THEN product = "Editor's Pick" | HIGH | | Complete server-rendered JSON-LD | Article + FAQPage + Speakable → AI-engine extraction/citation | MED | | Topical cluster + 15+ internal links | sibling posts on intent variants build topic authority | MED | | Immediate indexing (GSC + IndexNow/Bing) | catches the freshness window | MED |

The on-top workflow (end to end)

0. Research + EXISTENCE GATE: pick the query by demand × business-value × winnability; cluster by intent. HARD STOP — before writing ANY post, grep the content dir for an existing post on this query/intent (e.g. ls/grep src/features/blog/content). If one already exists, DO NOT write a new post — OPTIMIZE the existing one or skip. A duplicate cannibalizes both URLs. → keyword-research.md
1. Intent: open the live SERP; mirror the format the top 5 already use (usually listicle+table for "best X"). → content-onpage.md
2. Title: <N> Best <thing> in <YEAR> (<scope>). Year mandatory.
3. Draft for humans AND machines: answer-first intro, comparison table up top, one H2 per pick, honest trade-offs, product as "Editor's Pick", 5–6 FAQ Q&As, ~1,200+ words, self-contained quotable chunks. → content-onpage.md + geo-ai-engines.md
4. Internal links: 15+ to signup, pricing, /vs- pages, sibling + pillar posts.
5. Schema: Article (Person author, dateModified) + FAQPage + Breadcrumb + Speakable, server-rendered; site-wide Organization (+ real sameAs) + WebSite + SoftwareApplication. → schema-cookbook.md
6. Technical gate: renders in raw HTML, self-canonical, in sitemap w/ lastmod, hreflang reciprocal, not noindex, AI crawlers allowed, CWV not failing. → technical-seo.md
7. Index now: GSC "Request indexing" + IndexNow/Bing ping. → technical-seo.md
8. Entity: fill Organization.sameAs with real profiles; build off-page mentions. → entity-sameas.md
9. Measure + log: track GSC position + AI citations; after it moves, append a field-log entry and promote/demote levers. → measurement-iteration.md

Master pre-publish checklist

Content/GEO — [ ] intent matches SERP format · [ ] year in title · [ ] answer- first intro · [ ] comparison table up top · [ ] one H2 per pick · [ ] honest trade-offs + Editor's Pick · [ ] 5–6 FAQ (body + frontmatter) · [ ] self-contained quotable chunks · [ ] specific numbers/entities · [ ] 15+ internal links · [ ] cluster intent differentiated (no cannibalization).

Technical — [ ] renders in raw HTML (curl) incl. JSON-LD · [ ] self-canonical absolute · [ ] sitemap + accurate lastmod · [ ] hreflang reciprocal + x-default · [ ] not noindex · [ ] AI crawlers allowed (curl -A GPTBot) · [ ] CWV ok · [ ] GSC index request + IndexNow ping.

Schema/Entity — [ ] Article + FAQPage + Breadcrumb + Speakable, server-rendered · [ ] Organization + WebSite + SoftwareApplication site-wide · [ ] sameAs populated with resolving URLs · [ ] Person author for E-E-A-T · [ ] validated (Rich Results + curl | grep ld+json).

Highest-value open gaps (do next — see field-log)

• Organization.sameAs is empty → the #1 untapped entity signal. Populate with real LinkedIn/X/YouTube/Crunchbase/Wikidata/App Store URLs. → entity-sameas.md
• Article author is an Organization, not a Person → add founder Person for E-E-A-T.
• No Wikidata item yet → create one; durable entity anchor.
• 5 "best apps" sibling posts → watch GSC for cannibalization; differentiate intent.

Continuous-update rule (MANDATORY)

After ANY real ranking result tied to this playbook — a post hits/loses a position, an AI engine cites/drops the brand, a schema/sameAs change correlates with movement — append a dated entry to references/field-log.md using the template in references/measurement-iteration.md. Promote a hypothesis into "Proven levers" only after a logged win; demote/delete any lever the log later contradicts. The playbook MUST always reflect the log, not the other way around.