What happens when two of your pages answer the same question in AI search?
Usually one page gets picked and the other gets ignored, and you do not control which. Sometimes the weaker page wins. Sometimes the engine pulls a sentence from each and blends them. The fix is not to delete pages blindly. It is to decide which page owns each question and make the other page support it.
Overlap is normal on any site that publishes regularly. A blog post explains a concept. A service page explains it again. A glossary entry defines it a third time. Each page made sense when it was written. Together they compete for the same queries, and AI search makes that competition more visible.
I publish a lot, so I deal with this on my own site as much as on client sites. The principles below are the ones I use to decide what to merge, what to redirect, and what to keep.
How do AI search engines choose between similar pages?
AI search engines start from a retrieval step that finds candidate pages, then pick passages that answer the question best. For Google, the starting pool is its normal index. When two of your pages look alike, Google's own canonicalization process may treat one as the main version before any AI feature sees them.
Google's documentation on AI features says that "to be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet." So the usual indexing rules apply first. If a page is treated as a duplicate and not shown in Search, it is unlikely to be a supporting link either.
Google's canonicalization documentation explains what happens to near-duplicates: "If Google finds multiple pages that seem to be the same or the primary content very similar, it clusters them together." It then chooses the page it judges "most complete and useful for search users" as the canonical. You can influence that choice, but you do not get the final say.
Why is overlap a bigger problem for AI search than for classic search?
Overlap hurts more in AI search because the answer often cites only a few sources, and passages are judged on their own. A classic results page might show both of your pages. An AI answer usually shows one, and if the weaker page has the clearer sentence, that is the one that gets quoted.
Google says both AI Overviews and AI Mode "may use a 'query fan-out' technique," which it describes as "issuing multiple related searches across subtopics and data sources." That means a single question can trigger several related searches. Two of your pages might each match a different sub-question, which can be good. Or they might match the same sub-question, and then they split your chances.
The passage-level nature of AI answers is the key difference. I wrote about how engines choose specific sentences in how AI engines pick which sentence to quote. When two pages carry similar sentences, the engine has two near-identical candidates. Whichever is clearer, fresher, or better structured tends to win, regardless of which page you consider more important.
Is this the same as keyword cannibalization?
It is closely related but not identical. Keyword cannibalization is about two pages competing for the same ranking position. AI overlap is about two pages competing to supply the same answer passage. You can have AI overlap between pages that target different keywords, if both contain a strong answer to the same underlying question.
Classic cannibalization is usually found by looking at which URLs rank for a query in Google Search Console. I covered that process in how to fix keyword cannibalization on a Webflow site. AI overlap is harder to see because the competing pages may rank for different terms and still contain the same answer.
The practical difference is where you look. For cannibalization, look at queries. For AI overlap, look at questions. List the questions your pages answer, not the keywords they target, and you will find overlap you never noticed.
How do you find overlapping answers on your own site?
Build a question map. For each important page, write down the three to five questions it answers directly. Then sort the full list by question and look for repeats. Any question answered on two or more pages is a candidate for consolidation, especially if the pages give slightly different answers.
Different answers are worse than duplicate answers. If your pricing page says one thing about onboarding and a blog post says another, an AI engine might quote either. That is how brands end up with AI answers that contradict their own sales team.
For a small site, a spreadsheet is enough. For a large CMS, I export page titles, H2 headings, and first paragraphs into Airtable and group them by topic. Question-shaped headings make this much easier, which is one more reason to write H2s as real questions.
Which page should own the answer?
The page that owns the answer should be the one closest to the reader's next step. If the question is commercial, the service or product page should own it. If the question is educational, the guide should own it. Ownership means that page carries the full, clear answer and the other page links to it.
A common case is a definition that appears on both a glossary page and a service page. I usually let the glossary or guide own the full definition, and have the service page carry a short version that links across. The service page then focuses on its own job, which is explaining what you do and who it is for.
The reverse also happens. If a blog post answers "how much does this cost" better than your pricing page, the pricing page should be fixed to carry the answer properly. The blog post can then link to it and focus on the reasoning behind the price.
Should you merge, redirect, or rewrite overlapping pages?
Merge when two pages cover the same topic for the same reader. Redirect the weaker URL to the stronger one after merging. Rewrite when the pages serve different readers but share one answer, so each page answers its own question clearly and links to the other for the shared part.
Merging is the strongest move and the one people avoid, because it feels like losing content. In practice you keep the best paragraphs from both pages, put them on one URL, and send the old URL to it with a permanent redirect. Readers and crawlers both end up in one place.
Rewriting is the lighter option. Keep both pages, but change one so it no longer carries a complete answer to the shared question. Replace that section with a short summary and a link. This keeps both pages useful while removing the head-to-head competition.
Does adding special AI markup fix overlap?
No. Google states that "there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." Its documentation also says you do not need new machine-readable files or markup to appear in those features. Overlap is a content structure problem, and it needs a content structure fix.
Structured data still helps Google understand what a page is, and that has value. But no schema type tells an AI engine "prefer this page over my other page." Only clear ownership, internal links that point toward the owner, and canonical signals that agree with each other do that.
This is where teams often waste the most time. They add markup, files, and plugins hoping to steer AI answers, when the real issue is that their own pages disagree with each other.
What should you do next?
List the ten questions that matter most to your buyers. For each one, find every page on your site that answers it. Pick one owner per question, strengthen that page's answer, and either merge, redirect, or trim the competing pages so they link to the owner instead of repeating it.
Then check your internal links. Every page that mentions the topic should point to the owner with clear anchor text. That consistency tells search engines and readers alike which page speaks for you on that question.
If you want help building a question map for your site and deciding which pages should own which answers, reach out. Let's chat about your content.
Get found, cited and the back office automated
Let's make your site the source AI engines quote and wire up the systems behind it.
Read more blogs
Let's get your website found and cited by AI
Tell me what you're working on, whether AI search is skipping your product, your back office is buried in manual work, or you need a build that does both.