← all content

Week 4 of 5 · Class 8 of 10 · Thu 6/11

Be the source LLMs cite

When someone asks ChatGPT, Claude, or Perplexity a question about your industry, do you get named and linked? Today: warm up on a fresh model, then make one page quotable.

View the Class 8 deck on Gamma →

Pre-read · 5 min

Do this before class so Walk and Run land.

  1. The llms.txt argument is the wrong argument. Why llms.txt won't get you cited, and what actually does.
  2. Why Your Worst SEO Pages Could Be Your Best AEO Pages (advanced).

Rewatch video

Embed not loading? Watch on YouTube →

Today's schedule

4:00–4:15Fable 5 warm-up. Get cited by a brand-new model
4:15–4:30SEO 101 vs LLM SEO 101. How answer engines pick
4:30–4:45Walk. Why the cited pages win
4:45–5:05Run. Make one page quotable
5:05–5:25Team build. Week 4 teams
5:25–5:30Shareback

Warm-up: get cited by Fable 5

A brand-new Claude model just shipped, and no LLM has been trained on its launch yet. Can you get it to recommend the release when someone asks for the best option?

15 min
  1. Read the press release on why Fable 5 / Mythos 5 is awesome.
  2. Evaluate it against an ask for recommendations.
  3. Show the class a before / after of why this version is much better.

Recommendation queries to test

  1. Best dentist near me
  2. Best coding IDE for making enterprise apps
  3. Best way to test your product sense

Walk: why do the cited pages win?

Read the winners the way we read the Stripe docs and isitagentready.com last week. Open what an agent reads, and spot the pattern.

15 min
  1. Open the sources Perplexity, Claude, or ChatGPT cited for one query.e.g. best dentist near me · best coding IDE · best class for a product-sense interview. Why does it rank results that way?
  2. Prepend r.jina.ai/ to each to read what the agent reads.Does an agent-friendly llms.txt actually make the site rank higher?
  3. Spot the pattern: answer up top? title = the question? schema?

LLM SEO 101: how answer engines pick

  1. Retrieve. It grabs the paragraph that answers the question, not your whole page.
  2. Synthesize. It writes one answer quoting and summarizing just a few sources.
  3. Big and trusted ≠ cited. The clearest, most specific answer wins.

Run: make one page quotable

Ship the one change your Walk surfaced: atomize a page so an agent can lift the answer clean.

20 min
  1. Pick one real question your page should own.
  2. Make the H1 that exact question.
  3. Answer it in the first 40–60 words, detail below.
  4. Optional: have Claude write FAQ schema, validate in Google Rich Results Test.
  5. Re-ask your money question. Did it move? If it's too fresh to re-crawl, predict the lift.

Tools

Perplexity. Shows the sources it cited, inline.
Jina Reader. Prepend r.jina.ai/ to any URL to read what an agent reads.
Ahrefs authority checker. Score the cited page's authority.

Team challenge: can you hack the LLM ranking?

Same teams as Week 3. Pool your audits, find a gap in your space, and ship one citeable change together. Be ready to show a before vs. after.

20 min
  1. What's a gap in this space your team can own?
  2. Ship one atomized change together.
  3. Show a 60–90 second before vs. after.

What does done look like?

STRUCTURE. Can an agent lift a clean answer?
SPECIFICITY. Does one page own one question?
PROOF. Did the engines move?

Week 4 teams

1Nikoo Beyzaei + Chris Thorne + Daliso Zuze + Daniela Perilla
2Jeff Lash + Michael Krafft + Aviral Gupta + Jocelyn Baun
3Patricia + Sabrina Abhyankar + Kevin Ko + Elizabeth Ferrao
4John Dufresne + Ramin Talaie + Dorincy Shen + Arpit

Before you log off