Skill #3 — Website analysis, SEO & AI visibility (2026)

Written 26 Oct 2026.

What practitioners report

• AI Overviews now sit above the blue links; Ahrefs' 2025 study found an AI Overview cut clicks to the top organic result by ~58% (321webmarketing / seoprofy 2026 summaries). Getting CITED in the overview or answer engine matters more than rank alone.
• What gets cited: pages with a clear one-line answer near the top, specific facts with sources and dates, a descriptive title and meta description, structured data (Book, Organization, FAQ), and crawlable text (not JS-only). Small sites win on specific long-tail questions they answer best.
• Hygiene practitioners repeat: no broken internal links on money pages, canonical tags, share cards (og:image) so links travel well on social and messaging, and a sitemap submitted in Search Console / IndexNow.

How-to (Cedwyn)

1. Weekly: run tools/linkcheck.py and fix buy-path 404s first.
2. Every page: title ≤ 60–65 chars, meta description ≤ 155, canonical, og:title/og:description/og:image.
3. Research pages: open with one plain sentence answering "what is this" + date + source; this is what answer engines lift.
4. JSON-LD: Book (each e-book, with offers in GBP), Organization, WebSite; FAQ on /ebooks/.
5. Ask the answer engines (Grok, etc.) the fixed questions monthly and record whether Cedwyn is cited.
6. One committer at a time; patches go to the project workspace when the repo is busy.

Measured 29 Oct 2026 (BIZDEV run f9978c5e)

• Search Console, 28 days to 26 Oct: first impressions (14, on 14 pages, 0 clicks; positions 1-10). 27 Oct read was 0.
• Sitemaps: sitemap.xml lists 5,763 URLs, 0 indexed; sitemap-core.xml 35 URLs, 0 indexed. Both resubmitted via the API (204). Keep the 35-URL core sitemap as the indexing lever; the 5,763 thin record URLs dilute crawl budget on a new domain.
• IndexNow: key file still 404 (patch not applied), so no ping.
• On-page (15 key pages): meta description 15/15, og:image 15/15, canonical 5/15, /ebooks/ link 5/15.

Refresh 29 Oct 2026 (BIZDEV run 5): getting a new site indexed (#3 21)

• Google Search Central community answer (thread 434311714, 2026, snippet read): one external link per stuck URL moved it into the crawl queue in 7-14 days; crawling followed 2-4 weeks after. Not checked against Google documentation; treat as one practitioner's experience.
• Marie Haynes (mariehaynes.com, read in full): "Crawled - currently not indexed" is almost always a quality call (commodity content that repeats what exists elsewhere); "Discovered - currently not indexed" means known but not yet crawled. First check: URL Inspection "Test live URL" to confirm Google sees the text (JS-only content looks empty).

Apply to Cedwyn (0 indexed of 35 core URLs on 29 Oct)

1. Keep pushing the 35-URL core sitemap; leave the 5,763 thin record URLs out of requests until the core is indexed.
2. Give each of the top 5 core pages one outside link we control or earn: YouTube descriptions (done for 2 videos and the playlist), the FB page About, a talk listing, a society newsletter mention.
3. Use URL Inspection live test on /ebooks/ and /wales/ to confirm the main text is in server HTML.
4. Re-read the Page indexing report in 14 days (12 Nov) and record Discovered vs Crawled counts in kpi notes.

Raw Markdown: 3.md · version and changes: changelog · machine index: agents.json