SEO when search itself is getting weird.

• 01Ahrefs — keyword researchweekly
• 02Claude — drafts + topical clustersweekly
• 03Notion — content calendar + intent mapweekly
• 04Search Console — Google performancedaily
• 05LLM visibility tracker — Perplexity, ChatGPT, Claudeweekly

1. 0130 min

   Map your top 20 queries by intent.

   Two columns. Most teams discover their top 20 splits roughly 30/70 toward research. That's the budget reveal — you've been optimizing the wrong queries.

   Transactional

   Research

   Intent
   someone ready to buy or sign up

   Intent
   someone learning or comparing

   Engine
   Google still owns this

   Engine
   Perplexity, ChatGPT, Claude

   Query shape
   "buy [thing]", "[brand] pricing", "near us"

   Query shape
   "how does X work", "X vs Y", "best X in 2026"

   SEO play
   Page depth, internal links, schema, fast load

   SEO play
   Clean structure, direct claims, dated citations

   Win condition
   Click-through & convert

   Win condition
   Be the cited source
2. 02ongoing

   Transactional → traditional SEO.

   Page depth, internal links, schema, fast load. Standard moves. These pages are short and have one job — convert from search intent to action. Headline is the query, in declarative form.

   This funnel still works. It just got smaller.

3. 03ongoing

   Research → answer-engine SEO.

   Clean structure. Direct claims. Citable sources with publication dates. The page has to be parseable by an LLM that's going to summarize it for someone who'll never click through. Win condition is being the source, not the destination.

   Specific shapes that work: FAQ sections with question headers, comparison tables with named alternatives, "X vs Y" pages with explicit verdicts.

4. 04weekly

   Track LLM citations weekly.

   Ask Perplexity, ChatGPT, and Claude your top 10 research queries. Note which pages they cite. The list won't include all your top-ranking Google pages — that's the gap.

   Pages that rank in Google but never get cited in LLMs are the rewrite candidates.

5. 05quarterly

   Kill pages past page 2 with no LLM citations.

   If a page ranks below 20 in Google AND isn't cited by any LLM, it's tax. Either rewrite it as a citable source or 301 to a stronger page. Don't keep dead inventory.

   This is the move teams skip. It's also the one that recovers crawl budget and pushes the survivors up.

• ×Listicles. "Top 10" pages don't rank well anywhere and aren't cited by LLMs.
• ×Keyword stuffing. Reads as bot output. LLMs skip these in citations.
• ×3,000 words for a 200-word answer. Padding kills both rankings and citations.
• ×Backlink farms. Worked in 2018. Doesn't now.
• ×Ignoring zero-click queries. The answer-engine traffic happens there.
• ×Writing for Google alone. The half of your audience that uses Perplexity will never see those pages.

Two funnels now. Two content shapes. Stop optimizing for the funnel that's shrinking. Optimize for being cited, not just for being clicked.

Steal one thing: the intent map. Twenty queries, two columns. The shape of your content debt becomes obvious in 30 minutes.

FAQ, HowTo, Article schema. Same JSON-LD that's been around for years — now read by Perplexity and friends. Add it to every research page. Easy lift, real signal.

Every claim gets a source with a date. Not "studies show" — "Stanford HAI Report, March 2026, p.12." LLMs cite pages that cite back, because their reranking algorithms reward verifiability.

Stop building clusters around keywords. Build them around questions a real human actually types. Use AlsoAsked or just read Reddit. The cluster that earns citations is the one shaped like a conversation.

Get harder, but possible. UTM-link your own content when LLMs cite it. Watch Plausible for the referral. Track which LLM sends what. Pattern is still emerging — log early.