Skip to content
Chalk Labs / AI SEO
How-to guide

How to Get Cited in Google AI Overviews

By Rahil Jain, founder of Chalk LabsUpdated How we research

The short answer

To be cited in a Google AI Overview, a page must be indexed and eligible to show in Google Search with a snippet. Google says nothing more is required, including special schema or AI text files. Beyond that, answer the related sub-questions Google may search at the same time, write clear original passages, and track results in Search Console. A citation is never certain.

  • Google's only stated requirement: the page is indexed and eligible to show with a snippet. No special markup or AI files.
  • In one Ahrefs study, only 37.9% of cited URLs came from the top 10 in March 2026, so ranking well is not enough.
  • Cover the sub-questions around your topic, because Google may run several related searches at once.
  • Citations churn, and an AI Overview can cut clicks, so measure leads as well as impressions.

What must a page have to be cited in an AI Overview?

A page must be indexed and eligible to show in Google Search with a snippet. Google says that is the whole technical requirement, and that you need no new machine-readable files, AI text files, markup or special schema.org data 1.

Check three things for each page. It must be indexed, which Search Console's URL Inspection tool reports. It must be crawlable in robots.txt and by any CDN or hosting layer, which Google lists as a best practice 1. It must carry no snippet-blocking control. Google says nosnippet and max-snippet also limit what AI Overviews and AI Mode can use 14. Its pages do not say how data-nosnippet or noindex affect AI features.

Eligibility is only the entry ticket. Google describes its generative AI features as rooted in its core ranking and quality systems, so ordinary SEO best practice still applies 2. The script in the templates checks the basics for any URL.

Do it step by step

0 of 9 done0%

  1. Choose ten searches your buyers type. Run each in a logged-out browser window and note whether an AI Overview shows, which sites it cites, and whether the cited pages are articles, videos or forum threads. Screenshot each one with the date.

    Tip: Include at least three buyer-choice searches, such as 'best X for Y', because the AI Overview mix has shifted towards them 8.

  2. Run the eligibility script from the templates on every page you want cited, then paste the URL into Search Console's URL Inspection. You want a 200 status, no noindex or nosnippet, and Googlebot allowed. Fix blockers before you change any wording 1.

    Tip: Check the CDN too. Google lists crawl access at the CDN and hosting layer as part of the basics 1.

  3. For every search, list five to eight sub-questions. Take them from the AI Overview's own wording, the 'People also ask' box, your sales calls and the Fan-Out Explorer. Record them in the worksheet from the templates, with the page that should answer each.

    Tip: If one sub-question has no page, that is your first new section.

  4. Use the HTML skeleton from the templates. Give each sub-question an H2 phrased as a question, open with a one-to-three sentence answer, then add evidence. Keep the opening answer between 40 and 75 words so it can stand alone.

    Tip: Test your opening with the free Answer Block Checker.

  5. Add an original element to each key page: a price table with a checked date, your own photo, a measured result, a named expert with a one-line credential, or a short case told with permission. Google asks for content that is unique, compelling and useful 2.

    Tip: If you cannot add anything original, ask whether the page should exist.

  6. If the page carries JSON-LD, compare each value with the visible text. Delete anything a reader cannot see on the page, because Google's policy says markup must be a true representation of the content 5.

  7. After publishing, update the page's lastmod date in your sitemap only if the content really changed. Then use URL Inspection to request indexing for the most important pages. Allow a few days before you judge anything.

    Tip: Do not request indexing for dozens of pages a day. Prioritise the ten that matter.

  8. Open the Generative AI performance report and record impressions for your ten pages each week 3. Note any new sub-question that starts to appear. If the report is missing, your site may have too few impressions yet 3.

  9. Re-run the ten searches on the same day each month and compare with the first screenshots. Expect cited pages to change between runs 9. Keep a log of what you changed and which pages gained or lost citations.

    Tip: Judge the six-month trend, not any single month.

Your ticks are saved in this browser only.

Copy-ready templates

Python 3: eligibility check for one URL

text
#!/usr/bin/env python3
"""Usage: python3 aio_eligibility.py https://example.com/page

Checks the basics Google lists for appearing in AI Overviews: the page loads,
is not blocked from indexing or snippets, and is not disallowed in robots.txt."""
import re
import sys
import urllib.error
import urllib.request
from urllib.parse import urlsplit
from urllib.robotparser import RobotFileParser

url = sys.argv[1]
UA = "Mozilla/5.0 (compatible; eligibility-check)"


def get(u):
    req = urllib.request.Request(u, headers={"User-Agent": UA})
    try:
        with urllib.request.urlopen(req, timeout=20) as r:
            return r.status, r.headers, r.read().decode("utf-8", "replace"), r.geturl()
    except urllib.error.HTTPError as e:
        return e.code, e.headers, "", u
    except Exception as e:  # network error, bad host, timeout
        sys.exit(f"Could not fetch {u}: {e}")


status, headers, html, final = get(url)
print("Final URL:", final)
print("Status:", status, "(want 200)")

xrt = headers.get("X-Robots-Tag", "(none)")
print("X-Robots-Tag header:", xrt)

metas = re.findall(r"<meta\b[^>]*>", html, flags=re.I)
robots_meta = [m for m in metas if re.search(r"name\s*=\s*[\"']?(robots|googlebot)\b", m, flags=re.I)]
print("Robots meta tags:", robots_meta or "(none)")

blocked = re.search(r"noindex|nosnippet|max-snippet\s*:\s*0", " ".join(robots_meta) + " " + xrt, flags=re.I)
print("Blocks indexing or snippets:", "YES, fix this first" if blocked else "no")
print("data-nosnippet uses:", len(re.findall(r"data-nosnippet", html, flags=re.I)))

canon = re.search(r"<link\b[^>]*rel\s*=\s*[\"']?canonical[\"']?[^>]*>", html, flags=re.I)
print("Canonical tag:", canon.group(0) if canon else "(none)")

parts = urlsplit(final)
rstatus, _, rbody, _ = get(f"{parts.scheme}://{parts.netloc}/robots.txt")
print("robots.txt status:", rstatus)
if rstatus == 200:
    rp = RobotFileParser()
    rp.parse(rbody.splitlines())
    verdict = "allowed" if rp.can_fetch("Googlebot", final) else "BLOCKED"
    print("Googlebot per robots.txt:", verdict)
else:
    print("No usable robots.txt: crawlers may treat everything as allowed (4xx) or wait (5xx).")

Save as aio_eligibility.py and run python3 aio_eligibility.py https://yourdomain.com/page. It reports the status code, X-Robots-Tag, robots meta tags, data-nosnippet uses, the canonical tag and whether robots.txt lets Googlebot fetch the page. Python's robots parser is a rough check. Confirm indexing in Search Console, because this script cannot see Google's index.

HTML skeleton: one question per section, answer first

html
<article>
  <h1>How much does a new roof cost in [City]?</h1>

  <p class="answer">A new asphalt shingle roof in [City] typically costs
  $[low] to $[high] for a [size] home, checked on [date]. Metal costs more.
  Size, pitch, material and tear-off drive the price.</p>

  <h2>What affects the price of a new roof?</h2>
  <p>Four things move the price: size, pitch, material and the cost of
  removing the old roof. [One sentence on each, with a local number.]</p>

  <h2>Is replacing a roof cheaper than repairing it?</h2>
  <p>Repair costs less per job, but replacement wins when [rule of thumb].
  [Evidence: your own jobs, with permission, or a named source.]</p>

  <h2>How long does a new roof last?</h2>
  <p>[Answer in one sentence, then a table by material with a source.]</p>

  <p><small>Last checked on [date] by [name], [role].
  Prices come from [n] quotes we gave in [year].</small></p>
</article>

Replace every bracket. Keep the answers in plain visible text. If you add JSON-LD, every value in it must also be visible on this page.

Sub-question worksheet (copy into a sheet)

markdown
| Main search | Sub-question | Where it came from | Page that answers it | Original element to add | Done |
|---|---|---|---|---|---|
| how much does a new roof cost | roof cost per square foot | AI Overview text | /roof-cost | local price table | |
| how much does a new roof cost | roof replacement vs repair cost | People also ask | /roof-cost | two real job examples | |
| [your search] | [sub-question] | [source] | [URL or NEW] | [photo, data, quote] | |

Fill one row per sub-question. Any row marked NEW is a section or page to write.

Monthly AI Overview log

markdown
| Month | Search | AI Overview shown? | Domains cited | Are we cited? | Our impressions (Search Console) | What we changed |
|---|---|---|---|---|---|---|
| 2026-11 | [search 1] | yes / no | | yes / no | | |
| 2026-11 | [search 2] | yes / no | | yes / no | | |

Run it on the same day each month, logged out, from the same country. Screenshots are your evidence.

Do you have to rank in the top 10 to be cited?

Often not. In Ahrefs' March 2026 sample of 863,000 searches and 4 million cited URLs, 37.9% of cited URLs sat in the first 10 results, 31.2% in positions 11 to 100 and 31.0% beyond 100 4. Ahrefs suggests query fan-out may explain it: pages that appear most often across the sub-searches get cited 4. Google's documentation confirms the technique but not that explanation 1.

Ahrefs also quotes about 76% from an earlier July 2025 study, but says it changed its parsing since, so do not read 76% to 38% as an exact fall 4. All the figures come from Ahrefs' own data. The practical lesson holds either way. A page that ranks well still has a head start, but a page that ranks 40th for the main query can be cited for a sub-question it answers best.

How does query fan-out change what you should write?

Google says AI Overviews and AI Mode may use a 'query fan-out' technique, issuing multiple related searches across subtopics and data sources to build a response 1. Google's launch post for AI Mode says those searches run at the same time 7. So your page is judged against each sub-question, not only against the headline query.

Your job is to answer the sub-questions around a topic better than anyone else, on one page or a tight cluster. Ahrefs advises covering a topic from every angle and studying the fan-out queries assistants run 4. The free Query Fan-Out Explorer and how to do query fan-out research show how.

Worked example (modelled by us, not observed): one query, six likely sub-questions
  • Main query: how much does a new roof cost
  • Sub-question 1: average roof replacement cost by material
  • Sub-question 2: roof cost per square foot
  • Sub-question 3: what affects roof replacement price
  • Sub-question 4: roof replacement vs repair cost
  • Sub-question 5: how long does a roof replacement take
  • Sub-question 6: how long does a new roof last
  • Page plan: one H2 per sub-question, answered in its first sentence,
  • with a local price range and the date you checked it.

What does Google say you do not need to do?

Google's July 2026 guide lists several tactics you can skip. It says Google Search ignores llms.txt files, that structured data is not required for generative AI search, and that there is no ideal page length 2. The table pairs each with our reading.

What Google does ask for is content that is unique, compelling and useful, written from your own knowledge and experience, on a crawlable page 2. Google says creating such content will likely influence your presence in generative AI search more than any other suggestion in its guide 2. The steps below build on that.

Tactics Google's guide says to skip, and what to do instead
TacticWhat Google's guide says [s2]What to do instead
llms.txt or other AI text filesGoogle Search ignores them, so they neither help nor harm.Skip it for Google. See what llms.txt is for other tools.
Cutting content into tiny chunksThere is no requirement to break content into tiny pieces.Write clear sections for human readers.
Writing in a special style for AIYou do not need to write in a specific way for generative AI search.Write plainly and answer the question first.
Seeking inauthentic mentionsSuch mentions are not as helpful as they might seem.Earn real mentions from real work.
A page for every phrasing variantDoing this mainly to manipulate results violates the scaled content abuse policy.One strong page per topic. See scaled content abuse.

Does schema markup help you get cited?

Not according to Google, and we found no study that isolates a schema effect on AI Overview citations. Google says there is no special schema.org markup to add, and structured data is not required for generative AI search 12. It still recommends structured data for rich results 2.

If you add markup anyway, it must match the page. Google's policy says not to mark up content that readers cannot see, and the markup must be a true representation of the page content 5. Markup that disagrees with the page breaks that policy and gains you nothing.

So treat schema as housekeeping. Use the free Schema Generator if you want rich results, and read what schema markup is for the wider picture.

Which searches show an AI Overview in the first place?

Fewer than you might think, and the share moves. Semrush's study of more than 10 million keywords found AI Overviews on 6.49% of queries in January 2025, 24.61% in July and 15.69% in November 8.

The mix changed too. Informational queries were 91.3% of those that triggered an AI Overview in January 2025 and 57.1% by October, as commercial, transactional and navigational queries grew 8. Buyer questions about price and choice now meet AI Overviews more often than a year ago.

Some topics are decided by video and forums. In Ahrefs' September 2026 snapshot of more than 3 million US queries, YouTube held 22.9% of the mention share among the 50 most-cited AI Overview sources, as a share of the combined citations of those 50 sources and Reddit held 18.5% 6. If those sites fill the citations for your topic, a page alone may not be enough. Read how to optimize YouTube videos for AI Overviews.

How should you write a passage Google can use?

Write each section so it answers one question in its first sentence and still makes sense if lifted alone. Google's ranking systems include passage ranking, an AI system that identifies individual sections of a page 11. A long page can therefore win on a single passage.

Google also says there is no ideal page length and no need to chunk content for AI 2. Do not slice good prose into fragments. Use headings that match real questions, a short direct answer, then the evidence.

Add something a rival page lacks: your own price data, a photo of the work, a test result, a named expert. Google asks for content that is unique and useful, and warns against recycling what others have said 2. How to show first-hand experience in your content lists ways.

Do not rush to rewrite dates. In Ahrefs' study of 16.975 million cited URLs, the top-3 Google AI Overview citations averaged 1,432 days since publication, against 958 for ChatGPT citations 12. Ahrefs calls Google the surface least influenced by freshness, though its figures for organic results are inconsistent between text and table 12. Update a page when its facts change.

What did an AI assistant search for when we asked this question?

We gave a Claude agent with web search the prompt 'how to get my website cited in Google AI Overviews' on 20 October 2026, and it ran five distinct searches, each a different sub-question.

That is the fan-out idea in miniature, but one run on one engine is only a sample, and Google's own system may split the question differently.

How do you measure AI Overview citations?

Use the Generative AI performance report in Search Console, which counts impressions in AI Overviews and AI Mode together 3. How to measure AI Mode traffic covers its rollout and limits. AI feature traffic is also counted inside the normal Performance report under the Web search type 1.

Expect churn. Authoritas tracked 2,104 US desktop keywords that showed AI Overviews, in snapshots from August 2024 to January 2025, and rated the sample about 70% volatile over two to three months (a score, not a count of changed pages) 9. That is older data, so use it as a warning about stability, not as a current rate.

A citation is also not a visit. Ahrefs found that an AI Overview correlated with a 58% lower average click-through rate for the top-ranking page, comparing December 2023 with December 2025 across 300,000 keywords 10. That figure is modelled, correlational and desktop-only 10. Track leads and brand searches, not only clicks.

Which mistakes cost the most time?

Most wasted effort falls into the two mistakes below. Each one is cheap to avoid once you know it.

Two common mistakes when chasing AI Overview citations
MistakeWhy it hurtsFix
A leftover noindex or nosnippet rule from stagingThe page loses eligibility, because it must be indexed and snippet-eligible 1.Run the eligibility script on your ten key pages.
Counting a citation as a winTop-result CTR fell 58% where an AI Overview showed, in Ahrefs' modelled data 10.Track enquiries and branded searches.

When is this not worth chasing?

If no AI Overview shows for your buyers' searches, there is nothing to win yet. Search ten of your target phrases in a logged-out window first. If most show none, put the effort into classic ranking, which Google says its AI features are rooted in 2.

Skip it too if you limit snippets on purpose, for example on paywalled pages, because those controls limit what Search shows from the page 1. For a broader view, read how AI Overviews differ from AI Mode.

What an AI assistant searches when it answers this

We gave the question below to AI search tools and recorded the web searches they ran behind the scenes. This is called query fan-out. Each phrase is a page you could publish to be found and cited.

The prompt

“how to get my website cited in Google AI Overviews”

The searches it ran
  • how to get website cited in Google AI Overviews
  • Google AI Overviews citation study top ranking pages overlap
  • Google AI features guidance for site owners
  • schema markup AI Overviews does it help
  • AI Overviews query fan-out how it works

Observed October 20, 2026 using Claude web search. Claude agents were told to run 6 to 10 searches, so that count is our setting, not Claude's own choice. Assistants differ and change often, so treat this as a sample, not a rule. See how we collect it.

Frequently asked questions

Does ranking number one on Google secure a place in an AI Overview?

Not on its own. In Ahrefs' March 2026 sample, only 37.9% of AI Overview citations came from the first 10 results 4. Ranking helps a page get considered, but the system also searches sub-questions, and a page must answer them 14. Google's stated requirement is only that the page is indexed and eligible to show with a snippet 1.

Can I opt out of AI Overviews?

Google says nosnippet and max-snippet also limit what AI Overviews and AI Mode can use 14. Its pages do not say how data-nosnippet or noindex affect AI features. These controls apply to snippets in Search generally, so check the effect on your normal listings before you use them. Blocking Google-Extended does not do this job 13.

Does blocking Google-Extended remove me from AI Overviews?

No. Google says Google-Extended is a standalone token that manages use of content for Gemini training and grounding, and that it does not affect a site's inclusion in Google Search 13. AI Overviews are a Search feature, so use the snippet controls instead 1.

Do I need FAQ schema to be cited?

No. Google says there is no special schema.org markup to add for its generative AI features 1. If you add FAQ markup anyway, the questions and answers must be visible on the page 5. Write the FAQ for readers first, then mark up what is already there.

How many AI Overview citations should I expect?

No one can say. Citations churn: Authoritas rated its US desktop sample about 70% volatile over two to three months in older data, a score and not a count of changed pages 9. Track the trend in Search Console's report 3 and in a fixed monthly log of your own searches.

About the author

Rahil Jain

Rahil Jain is an AI SEO (answer engine and generative engine optimisation) consultant and co-founder of Chalk Labs.

LinkedIn · Method and credentials

Sources

  1. AI features and your website · Google Search CentralOpened October 20, 2026. Supports: Pages must be indexed and snippet-eligible to appear; no extra technical requirements, AI text files or special schema; query fan-out technique; snippet controls; AI traffic counted in the Performance report under Web; crawl access in robots.txt and CDN. Page last updated 2025-12-10.
  2. Optimizing your website for generative AI features on Google Search · Google Search CentralOpened October 20, 2026. Supports: Google Search ignores llms.txt; no need to chunk content or write a special way; no ideal page length; inauthentic mentions not helpful; scaled content abuse policy; unique, compelling, useful content; AI features rooted in core ranking systems. Page last updated 2026-07-10.
  3. Generative AI performance report (Search) · Google Search Console HelpOpened October 20, 2026. Supports: Impressions in AI Overviews and AI Mode by page, country, date and device; rolled out to all websites worldwide as of 31 August 2026; may not appear with too few impressions; newest data preliminary.
  4. Update: 38% of AI Overview Citations Pull From The Top 10 · AhrefsOpened October 20, 2026. Supports: Published 2 March 2026: 863,000 SERPs and 4 million AI Overview URLs; 37.9% of cited URLs in the top 10, 31.2% in positions 11 to 100, 31.0% beyond; about 76% in July 2025 versus about 38% now; Ahrefs says its parsing changed since July 2025 and suggests query fan-out may explain the shift.
  5. General structured data guidelines · Google Search CentralOpened October 20, 2026. Supports: Do not mark up content that is not visible to readers; structured data must be a true representation of the page content.
  6. Most cited domains in Google AI Overviews (September 2026) · AhrefsOpened October 20, 2026. Supports: Over 3 million US queries: YouTube 22.9% and Reddit 18.5% mention share among the top 50 most-cited sources; mention share defined as a domain's citations as a percentage of the summed citations of those sources.
  7. Expanding AI Overviews and introducing AI Mode · Google (The Keyword blog)Opened October 20, 2026. Supports: Published 5 March 2025: AI Mode uses a query fan-out technique, issuing multiple related searches concurrently across subtopics.
  8. Semrush AI Overviews study · SemrushOpened October 20, 2026. Supports: AI Overviews on 6.49% of queries in January 2025, 24.61% in July and 15.69% in November (10M+ keywords); informational share of triggering queries fell from 91.3% in January to 57.1% in October 2025.
  9. SERP, organic and AI Overview volatility research · AuthoritasOpened October 20, 2026. Supports: Published 19 February 2025: 11,203 US desktop keywords, 2,104 with an AI Overview, snapshots in August 2024, October 2024 and January 2025; conclusion that about 70% of pages in AI Overviews change over two to three months.
  10. AI Overviews reduce clicks by 58% (update) · AhrefsOpened October 20, 2026. Supports: Published 4 February 2026: across 300,000 keywords, an AI Overview correlated with a 58% lower average click-through rate for the top-ranking page, December 2023 versus December 2025; modelled, desktop-only.
  11. A guide to Google Search ranking systems · Google Search CentralOpened October 20, 2026. Supports: Passage ranking is an AI system used to identify individual sections or passages of a web page. Page last updated 2025-12-10.
  12. Do AI Assistants Prefer to Cite Fresh Content? · AhrefsOpened October 20, 2026. Supports: Published 28 July 2025: 16.975 million cited URLs; Google AI Overview top-3 citations averaged 1,432 days since publication against 958 for ChatGPT citations; the author calls Google the least influenced by freshness.
  13. Google's common crawlers · Google Search CentralOpened October 20, 2026. Supports: Google-Extended is a standalone product token for managing use of content for Gemini training and grounding; it does not impact a site's inclusion in Google Search and has no separate user agent string. Page last updated 2026-10-15.
  14. Robots meta tag, data-nosnippet, and X-Robots-Tag specifications · Google Search CentralOpened October 20, 2026. Supports: nosnippet prevents content being used as a direct input for AI Overviews and AI Mode; max-snippet limits how much may be used; the page does not say how data-nosnippet or noindex affect AI features.
How this page was made. Built on Rahil Jain's AI SEO method and drafted with AI research assistance. Every statistic links to the page we read it on (opened October 20, 2026 or nearby). Published October 20, 2026; last updated October 20, 2026. Chalk Labs sells AI SEO services, so we have a commercial interest in this topic. Spotted an error? Tell us and we will fix it.