No one can guarantee inclusion in Google AI Overviews. But the signals AI Overviews use to pick sources are knowable, and you can improve every one of them. Below is the practical playbook: eight changes that compound, in priority order.

How AI Overviews pick sources

AI Overviews do not have their own ranking algorithm. They run on the same retrieval engine as classic Google results — fetching a small candidate set, usually drawn from page 1 — then a separate model re-ranks those candidates for excerptability. Three consequences:

  • If you are not on page 1 for the query, you are extremely unlikely to be in the Overview
  • Being on page 1 is necessary but not sufficient; the re-rank step is where the leverage now sits
  • The re-rank rewards clear structure, short quotable passages, and trust signals

1. Answer the query clearly in the first paragraph

The single highest-leverage change. Open the page with a one-paragraph direct answer to the query, written in the same words the searcher used. Then go into the depth. This is the paragraph Overviews will lift verbatim if you are cited; it is also what makes the page feel useful to actual humans.

2. Build topical depth around the question

One article about X loses to ten interlinked articles about X. Overviews pull from sources where the surrounding content shows topical authority — adjacent articles addressing sub-questions and objections, linked from the main page.

Pick a pillar topic. List 8-15 related questions buyers ask about it. Build one page per question. Link them all back to the pillar page. You will not regret it.

3. Use structured headings and FAQ sections

  • One H1 stating the question or topic
  • H2s for each sub-question, written as the user would phrase it
  • FAQ section at the end with 4-8 questions and short, direct answers
  • FAQPage JSON-LD reflecting those FAQ questions exactly

Overviews preferentially pull from FAQ structures because they are already in question-answer format — easy to excerpt with attribution. The schema doubles the signal.

4. Show E-E-A-T signals

Experience, Expertise, Authoritativeness, Trustworthiness. Google has been explicit that AI Overviews lean harder on these than classic rankings:

  • Visible author byline with credentials or relevant experience
  • Author page linked from each article, with bio and external presence
  • About page that says who you are and why you should be trusted on this topic
  • Citations to authoritative sources for any claim that needs one
  • Date of original publication AND last meaningful update, visible and in JSON-LD

5. Keep technical SEO clean

Sloppy technicals filter you out before the re-rank even runs. Specifically check:

  • The page is crawlable, indexable, and returns 200
  • The full content is in the raw HTTP response, not loaded by JS
  • LCP under 2.5s on mobile
  • Schema matches the visible page content (no faking FAQ schema)
  • Page is in your XML sitemap with a fresh lastmod

6. Use schema accurately

Schema is a hint, not a guarantee, but it improves the precision of how AI engines understand the page. For Overview-targeted content:

  • Article schema (or NewsArticle / TechArticle as appropriate)
  • FAQPage where there is a genuine FAQ section
  • HowTo where the page is a step-by-step guide
  • BreadcrumbList for site navigation context
  • Author schema linked to a Person entity

7. Refresh pages as queries evolve

AI Overviews weight freshness heavily — both publication date and meaningful updates. A page that ranked 18 months ago and has not been touched will quietly slip out of the candidate set. Re-audit priority pages quarterly:

  • Is the top paragraph still the best direct answer?
  • Are the H2s still aligned with current sub-question phrasing?
  • Has anything in your data, examples, or recommendations changed?
  • Are the cited sources still authoritative?

8. Measure both classic rankings AND AI visibility

Classic rank trackers do not capture Overview presence. Add an explicit check: monthly, query a representative set of prompts against Google AI Overviews and record whether your site is cited, mentioned, or absent. Tools like Profound, Otterly, and manual spreadsheet checks all work. The point is to have a baseline you can track.