AC
SEO & AEO4.6 KBMIT licensed

claude-seo-sxo

Rewritten from patterns in AgriciDaniel/claude-seo (MIT)

Search Experience Optimization, reads the SERP backwards to find page-type mismatches, derive user stories, and score pages per persona. Use when the user says "SXO", "why isn't my page ranking", "intent mismatch", "SERP analysis", "persona scoring", or "page type mismatch". Not for technical health checks: use claude-seo-technical. Not for full-site audits: use claude-seo.

  • claude
  • seo
  • sxo

SKILL.md

Search Experience Optimization (SXO)

A page can be technically perfect and still never rank because it is the wrong page type for the keyword. If the SERP shows eight product pages, your blog post loses no matter how well-optimized it is. Read the SERP backwards first, then judge the page.

Workflow

  1. Fetch the target. Retrieve the URL, extract title, H1, meta description, heading hierarchy, word count, schema, CTAs, and media. If the page is JS-rendered, work with the available HTML and note the limitation. If no keyword given, derive the primary keyword from title-H1 overlap. No keyword derivable: ask, do not guess.
  2. Read the SERP backwards. Search the keyword. For each top-10 organic result, record domain tier, page type, content format, depth estimate, and schema signals. Record SERP features: featured snippet format, People Also Ask questions, ad copy themes, related searches, AI Overview presence. Compute consensus: dominant page type (>60% strong, 40-60% mixed, <40% fragmented), depth norm, schema expectation, media expectation.
  3. Detect page-type mismatch. Compare target type against consensus using the severity table in references/sxo-method.md. Mismatch is the number one finding, report it before anything else. A fragmented SERP is a differentiation opportunity, say so.
  4. Derive user stories. PAA reveals knowledge gaps, ad copy reveals commercial triggers, related searches reveal the journey, snippet format reveals the expected answer shape. Write 3-5 stories: "As a [persona], I want [goal], because [driver], but I'm blocked by [barrier]." Cite the SERP signal behind each story.
  5. Score the gap. Score 7 dimensions (page type, depth, UX, schema, media, authority, freshness) per the rubric in references/sxo-method.md. Output a 0-100 SXO Gap Score. Keep it separate from any technical SEO score, they measure different things.
  6. Score per persona. Derive 4-7 personas from the SERP signals. Score each on Relevance, Clarity, Trust, Action (25 points each). Rank recommendations weakest persona first, that is the biggest opportunity.
  7. Prioritize. Fix the type mismatch first, then the weakest persona gaps, then depth and media. End with an honest limitations section: sample size, data source, what you could not assess.

Hand off findings: schema gaps to claude-seo-schema, technical issues found during fetch to claude-seo-technical.

Verification

Do this before delivering: count the SERP results you actually analyzed. Expect at least 5 with page type classified for each. Fewer than 5: state the limited sample in the Limitations section, do not silently pretend full coverage. Then check every user story cites a specific SERP signal. Any story without a cited signal: rewrite it or cut it.

Good vs Bad

Bad: "Your page needs more content and better CTAs." Generic, ignores the SERP, no evidence.

Good: "SERP consensus is 7/10 comparison pages. Your URL is a single-product landing page: CRITICAL mismatch. Restructure as a comparison with a feature matrix, or build a separate comparison page. Then fix the 'skeptical evaluator' persona (score 40/100): no third-party proof above the fold."

Footguns

  • Scoring the page before reading the SERP. You will optimize the wrong page type. Fix: SERP analysis is always step 2, never skipped.
  • Inventing personas from imagination. Personas must come from PAA, ads, and related searches, or they are fiction. Fix: cite the signal for each persona.
  • Blending SXO score into technical score. A 95 technical / 30 SXO page reads as "fine" when averaged. Fix: report both scores separately, always.
  • Vague wireframe advice. "Add a CTA here" is useless. Fix: ultra-concrete placeholders, "add pricing CTA with annual-savings badge below hero, linking to /pricing".

Completion checklist

  • [ ] At least 5 SERP results classified by page type
  • [ ] Mismatch verdict stated with severity
  • [ ] 3-5 user stories, each citing a SERP signal
  • [ ] Gap score and persona scores reported, separate from technical score
  • [ ] Priority actions ranked, limitations stated honestly

Any box unchecked: not done. Fix or say so.

Reference files

More in SEO & AEO

ai-seoOptimize content to be cited by AI search engines and assistants (AI Overviews, ChatGPT, Perplexity, Claude, Gemini, Cop...claude-seoSEO analysis orchestrator, routes audits to the right specialist skill and merges results into one scored report. Use wh...claude-seo-backlinksBacklink profile analysis: referring domains, anchor text distribution, toxic link detection, competitor link gap. Use w...
All skills