Why Does Google AI Overviews Cite a Forum Answer Instead of Our Official Product Documentation?

Why does the forum answer win?

• It uses the user's words. A search for "sync stuck at 99% after update" matches a thread with that title. Your docs page is called "Configuring synchronisation settings".
• It contains the error text. People paste error messages into search. Forum threads quote them, and documentation rarely does.
• It answers directly. "Clear the cache folder and restart; fixed in 4.2" is exactly the passage an AI summary wants.
• It names versions and environments, such as the operating system, browser or plan, which match specific searches.
• It is recent. A thread from last month can look fresher than documentation nobody has touched in two years.
• Google surfaces discussions. Google supports markup that helps it identify forum discussions for its "Discussions and forums" feature, and forum pages rank well for problem-style searches.
• Ranking is not everything. Ahrefs' 2026 analysis of AI Overview citations found only 38% of cited URLs ranked in the top ten for the query, and 31% ranked beyond position 100. AI Overviews pick passages that answer parts of the question, so a well-matched passage can win from a low position.

What stops documentation being cited?

• Script-rendered documentation portals that show little text in the page source.
• Login walls or cookie gates on some or all docs.
• Version sprawl: the same page under many version URLs, with unclear canonical tags.
• Fragmentation: the answer spread across a reference page, a changelog and a release note. Why fragmentation causes wrong answers is covered in how fragmented documentation and changelogs cause AI to give wrong answers.
• No troubleshooting content: docs explain how things should work, not what to do when they do not.
• Blocked or noindexed docs subdomains, often left over from a migration.

Does it matter if a forum is cited?

It depends on the answer:

Sending feedback on an AI Overview is covered in what to write in a Google AI Overview report-a-problem form.

What should you change in your documentation?

1. Write troubleshooting pages keyed to symptoms. Title them the way users describe problems: "Sync stuck at 99%", "Login loop after password reset".
2. Include exact error text in the page, as text, not only in screenshots.
3. Put the fix first, then the explanation. State which versions and environments it applies to.
4. Keep a "known issues" page for each release, with workarounds and fix versions.
5. Serve docs as plain HTML, readable without scripts, with no login for public documentation.
6. Use one stable URL for the current version of each page, with older versions clearly marked and canonicalised.
7. Show when each page was most recently updated, and review high-traffic pages on a schedule.
8. Use search data. Your docs site's internal search and Search Console queries show the words people use. Turn frequent problem queries into troubleshooting pages.

What should you do in forums?

• Answer in your own community with staff accounts, link to the official page, and mark accepted solutions. If you run the forum, add DiscussionForumPosting markup to open discussions, or QAPage markup to threads that are a question with answers, so Google can identify posts and answers clearly.
• Where third-party forums allow it, post a helpful reply from an identified company account, linking to the official fix without marketing.
• Turn recurring forum questions into docs. If the same question appears every week, the docs are missing a page.

Does the same happen in other assistants?

Yes. ChatGPT, Perplexity and Copilot also cite forum threads for troubleshooting questions, for the same reasons: matching words, direct fixes and recent dates. Documentation portals that render by script fare even worse there, because most AI crawlers other than Google's do not run JavaScript, so they may see an almost empty page. Check your robots.txt and any bot-protection rules on the docs subdomain, and make sure the search crawlers of the assistants your users rely on can fetch public documentation. The same troubleshooting pages then serve every assistant, not just Google.

How do you check progress?

List the problem-style questions where forums are cited. Search each one in Google and note the AI Overview's citations, then repeat every two weeks after publishing the new pages. Also check Search Console for impressions on the new troubleshooting pages, and the generative AI performance report for impressions by page in AI Overviews and AI Mode, which the report combines. Finding which pages AI engines cite for target questions is covered in how to find which pages AI answer engines cite for your target questions.

What might an illustrative case look like?

Consider an illustrative accounting software company whose AI Overview for "bank feed not updating [product]" cited a three-year-old forum thread. The thread recommended reconnecting the bank, which no longer worked after an API change. The official docs had a page called "Managing bank connections", rendered by script, with no mention of the error message users saw. The company published a troubleshooting page titled with the exact symptom and the error text, with the current fix in its first lines and the affected versions. It also added a staff reply to its own community thread linking to it. Within a month, the AI Overview cited the new page, and support tickets on the issue fell.

What do people ask about forums and AI Overviews?

Should we stop people discussing our product in forums?

No. Forums are where users help each other. Make sure the official answer is easy to find and link to.

Does forum markup help our own community?

It helps Google identify discussions and posts on your forum. The content of the answers still matters most.

Why is the forum cited even though our docs rank higher?

AI Overviews select passages that answer the question, not just the top-ranking page. A passage that matches the user's words and gives a direct fix can win.