Cristhianzl/claude-skills-czl / configs/agnostic/skills/optimizing-discoverability
optimizing-discoverability
VerifiedOptimizing discoverability — SEO · AEO · GEO
Three lenses on one goal, each aiming at a different "searcher": SEO wants the click on a results page; AEO wants to be the answer (featured snippet, People Also Ask, voice); GEO wants to be the source an AI cites (ChatGPT, Claude, Perplexity, AI Overviews). They compound: answer engines and generative engines overwhelmingly pull from pages that already index and rank well — SEO is the foundation; AEO and GEO are layers on top, not replacements.
Read first (always)
List learnings/ and read anything relevant — the project's niche, existing rankings, target queries, analytics stack, and past decisions live there and override the defaults here.
Grounding — what this skill is based on
- Google's official AI-features guidance (developers.google.com/search/docs/fundamentals/ai-optimization-guide): AI features "are rooted in our core Search ranking and quality systems" — pages must be indexed and snippet-eligible; the winning content is "unique, non-commodity" with a real point of view.
- The GEO study (Aggarwal et al., KDD 2024 — arxiv.org/abs/2311.09735): first large-scale test (9 tactics, 10k queries) of what makes generative engines cite a page. Adding citations to sources, quotations, and statistics boosted visibility 30–40%; keyword stuffing made it worse.
Myths — officially debunked, don't spend effort here
Google states these explicitly (for Google surfaces):
llms.txtand "AI text files" are ignored by Google Search. (For non-Google engines it's an emerging, optional standard — seereferences/geo.md; never treat it as the strategy.)- No chunking required — you don't need to fragment content "for the AI".
- No special schema is required for AI features; structured data clarifies what's already visible, it doesn't unlock anything hidden.
- Keyword density / long-tail stuffing is obsolete — systems understand synonyms and meaning; the GEO study measured stuffing reducing AI visibility.
- Artificial mentions/link schemes remain counterproductive.
Workflow
- Scope and goal. What does "found" mean for this project — clicks (SEO), the answer box (AEO), AI citations (GEO), or all three? Which pages, which market/language, what's the primary conversion? One measurable target.
- Audit what exists. Indexability first (robots, sitemap, canonicals, the technical floor in
building-frontend-ui/references/designing-from-scratch.md); then current rankings/queries (Search Console), current AI visibility (ask the major assistants the money questions — is the brand cited?), and the competitor set. - Map queries to intent to pages. Every target query gets one owning page and an intent label (informational / commercial / transactional / navigational). One page per intent cluster — never two pages competing for the same query.
- Apply the three passes, in order:
- SEO pass —
references/seo.md: intent match, E-E-A-T with real experience, topical authority, internal linking, technical health. - AEO pass —
references/aeo.md: question-shaped headings, a 40–60-word direct answer immediately under each, lists/tables for steps and comparisons, schema that mirrors visible content. - GEO pass —
references/geo.md: citable self-contained passages, named sources + statistics + quotations, original data, entity consistency, AI-crawler access policy.
- SEO pass —
- Measure and iterate. Search Console (clicks, impressions, position), snippet/PAA wins, AI-referral traffic and brand citations in assistant answers. Expect 3–6 months for meaningful movement; report that honestly instead of promising fast wins.
Non-negotiables
- Never fabricate E-E-A-T: no fake authors, fake reviews, invented statistics, or "as seen in" logos that aren't real. That's the discoverability version of the fake-testimonial rule (
building-frontend-ui/references/anti-vibe-coded.md) — and quality raters and AI engines are specifically tuned to smell it. - Content answers the query it targets — the promise in the title is kept in the first screen. Clickbait that under-delivers dies in every one of the three lenses.
- Schema mirrors visible content only. Marking up content that isn't on the page is a spam signal.
- The technical floor comes first. A page that isn't crawlable/indexable/snippet-eligible can't win any of the three games (floor:
building-frontend-ui). - Don't sacrifice readers for engines. Google's own summary: "Focus on what your visitors would enjoy, find helpful, and feel satisfied with."
Output
For an audit: a prioritized report — indexability blockers first, then per-lens gaps with the specific page and fix, then the measurement plan. For content work: the intent map, then the content with the three passes applied, then what to measure. State assumptions (market, language, competitor set) explicitly.
Capture a learning
Record the project's niche, target queries, what ranked/got cited and what didn't, and any engine-specific behavior you observe (learnings/YYYY-MM-DD-slug.md or /learn).
See also
references/seo.md·references/aeo.md·references/geo.mdskills/building-frontend-ui— the technical on-page floor (head, semantics, CWV) andreferences/designing-from-scratch.md(SEO foundation checklist).skills/writing-prd— when the real problem is "what content should exist", scope it before optimizing it.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |