Publish where AI assistants cite sources (GEO workflow)

Why this matters

Log which pages AI assistants cite for buyer questions, then publish that research on your own site.

Before you start

• Access to at least two AI assistants (for example ChatGPT, Perplexity, Gemini)
• A list of the buying questions customers actually ask
• Ability to publish or edit pages on your own site

How to apply it

1. List the buying questions customers use

   Collect the questions buyers ask before purchase from sales calls, support tickets, site search and paid-search terms. Write them the way a customer phrases them, not as short head terms.

   Check: A document lists at least five full questions and where each was observed.

2. Ask each question in several AI assistants

   Run every question through more than one assistant and open the citations attached to each answer. Record the exact cited URL, the assistant, the date and the position in the answer.

   Check: A spreadsheet has one row per question and assistant with a full cited URL, not just the brand name.

3. Classify the cited sources

   Tag every recorded citation with the domain type (own site, forum, review platform, video, publisher), the page format and the topic it covers.

   Check: Every citation row carries a domain type, a page format and a topic tag.

4. Publish the full research on your own site

   Turn the topic into a first-party guide that carries the original research, data and evidence rather than a summary of the cited pages.

   Check: The guide URL returns HTTP 200, contains the original data, and its rel=canonical points to itself.

5. Adapt the guide into the formats you found

   Produce the assets that matched the citation patterns: a YouTube walkthrough, an answer in a relevant discussion, an X article, a LinkedIn post or a review-platform entry.

   Check: Each adapted asset contains a link to the first-party guide URL.

6. Keep product details current and link to the evidence

   Update prices, features and availability on the guide as the product changes, and add outbound links to the primary sources the research relies on.

   Check: Prices and feature statements match the live product pages and every outbound evidence link returns HTTP 200.

7. Separate AI referrals in analytics General best practice

   Filter sessions by the referrers used by AI assistants so traffic from those answers can be compared with search traffic.

   Check: A report filtered on the referrer pattern returns sessions in the last 28 days.

8. Repeat weekly and track outcomes

   Re-run the same question set on a weekly cadence, log new and lost citations, and keep a separate column for the trials or sales the cited pages produce.

   Check: The weekly log shows a cited-URL count per assistant per week plus an outcome column populated for at least one period.

<goal>
image sitemap anchor breadcrumb cluster markup priority breadcrumb measure anchor crawler image canonical image.
</goal>

<context>
pipeline snippet locale intent signal payload render heading cluster page pipeline payload anchor render.
</context>

<discover_first>
locale canonical render markup route link pipeline locale layout cluster serp page breadcrumb priority.
</discover_first>

<constraints>
breadcrumb index render page redirect snippet intent measure link canonical intent template serp metadata.
</constraints>

<plan>
verify sitemap route snippet heading sitemap report priority report heading section layout intent index.
</plan>

<verify>
effort template page heading priority render sitemap page template signal priority crawler link caching.
</verify>

<report>
link payload crawl redirect heading signal pipeline snippet signal query locale section sitemap route.
</report>

Source: @borjafat on 2026-09-28 · original post · report an error