If you have been watching your Google Search Console data lately, you may have noticed something odd. Some sites appear regularly in AI Overviews but get no mentions in AI Mode. Others are cited constantly in AI Mode conversations but barely show up in the traditional AI Overview snippets. These are not random quirks. The two systems are built differently, trained to do different things, and pull from sources using different logic. Understanding that difference is the first step to actually doing something about it.
What Each System Is Actually Doing
AI Overviews, formerly Search Generative Experience, sits at the top of a standard Google search results page. It generates a short summary answer to a query, then attaches a small cluster of source links, usually three to five. The job of AI Overviews is to give the searcher a quick, confident answer without requiring them to click anywhere. It draws heavily from pages Google has already ranked highly for that query. If you are ranking on page one, you have a realistic shot at being cited.
AI Mode is a different beast entirely. When a user switches into AI Mode, they are entering a conversational interface. Queries get longer, more complex, and often multi-part. The system reasons across multiple sources before forming a response, and it continues to do so as the conversation evolves. It does not simply summarise the top-ranked pages. It draws from a broader, more dynamic pool and weighs sources based on how well they answer the specific phrasing of the question being asked at that moment.
Think of AI Overviews as a quick reference tool and AI Mode as a research assistant. One skims the top shelf. The other digs through the whole library.
The Core Reason Citations Differ
The citation difference comes down to three things: query intent, source diversity, and the way each system evaluates relevance.
Query intent and depth
AI Overviews almost always handles short, definitive queries. "What is the capital of France?" "Best running shoes for flat feet." "How long does concrete take to set?" The answers are short and the sources need to be authoritative and concise. Google leans on its existing ranking signals heavily here. A page that ranks position one or two for a keyword has a very good chance of being pulled into the AI Overview for that same keyword.
AI Mode handles multi-step reasoning. A user might ask: "I run a small Shopify store selling handmade ceramics, what marketing channels should I focus on first with a budget of around £500 a month?" That is not a one-page answer. The system will synthesise from several sources, weighting each based on how closely it addresses a specific piece of the question. A page that ranks well for "Shopify marketing channels" might not appear at all if it does not also speak to budget constraints or small-scale operations specifically.
Source diversity
AI Overviews are conservative with citations. Three to five links, often from established domains, often from the same cluster of already-ranking pages. AI Mode is far more willing to pull from niche sources, mid-authority sites, and pages that sit outside the top ten for any given keyword, provided those pages answer a specific part of the conversation with precision.
This is actually good news for smaller brands. A well-structured page that answers a very specific question in a thorough, credible way has a better chance of appearing in AI Mode than in a traditional AI Overview. The bar for citation is different, not necessarily higher.
Structured data and page clarity
Both systems benefit from clear, well-structured content, but AI Mode is more sensitive to it. When a page uses proper schema markup, the model can parse the page's meaning with far greater confidence. A product page with correct structured data tells the model exactly what is being sold, at what price, to whom, and why. A blog post with FAQSchema tells the model there are discrete question-and-answer pairs worth extracting. Without that structure, the model is guessing at meaning from raw HTML, and guessing means you get cited less.
This is the core of what FlinnSchema works on: making pages machine-readable enough that AI systems stop guessing and start citing. You can see the difference this makes across different page types in our client results and case studies.
How Google Chooses Which Pages to Cite in Each Mode
For AI Overviews
Google's AI Overviews system relies heavily on the traditional web index and existing PageRank-style signals. If your domain has strong topical authority, your pages rank well, and your content is concise and factually grounded, you are a strong candidate. The system also pays attention to E-E-A-T signals, Experience, Expertise, Authoritativeness, and Trustworthiness. Author bios matter. External mentions matter. The age and consistency of your domain's topical coverage matters.
One important nuance: AI Overviews tends to prefer pages that match the query almost exactly. If someone searches "how to remove limescale from a kettle", a page titled "How to Remove Limescale From a Kettle" with a clear step-by-step structure is going to win over a 4,000-word article on descaling appliances that covers the kettle question in paragraph six.
For AI Mode
AI Mode operates more like a retrieval-augmented generation system. It pulls candidate pages, scores them against specific parts of the query, and then synthesises. This means a single long-form page can be cited for one section but ignored for another. It also means a page does not need to be the definitive authority on a topic. It just needs to be clearly, confidently correct on the specific sub-question it addresses.
Pages that tend to perform well in AI Mode share a few characteristics: they use clear headings that match the language of real questions, they include concrete specifics like numbers, timelines, prices, and named processes, and they avoid vague generalities that the model cannot confidently extract and repeat. Schema markup amplifies all of this. It is not magic, but it gives the model permission to trust what the page says.
If you are not sure whether your pages are set up to be cited in either environment, a free AI visibility audit is a sensible starting point. It shows you exactly where structure is missing and which pages have the most citation potential.
Practical Differences in What Gets Cited
Here is a concrete example. Imagine two businesses: a financial planning firm and a recipe blog.
The financial planning firm has strong domain authority, ranks in the top five for several high-volume keywords, and has been publishing content for eight years. They appear regularly in AI Overviews. But their pages are dense, lightly structured, and written in a way that buries specific answers inside long prose. When users ask complex, multi-part financial planning questions in AI Mode, the firm barely gets a mention. The model cannot quickly extract a clean, citable answer.
The recipe blog, meanwhile, has a relatively modest domain authority but uses structured Recipe schema on every post, includes clear timing information, ingredient lists, and FAQ sections under each recipe. They do not appear in AI Overviews for competitive queries, but they get cited constantly in AI Mode when users ask follow-up questions like "what can I substitute for buttermilk in that recipe?" The structure makes the pages trivially easy to mine for specific answers.
The lesson: AI Overviews rewards authority and ranking position. AI Mode rewards specificity and structure. Ideally, you build for both.
What to Do if You Are Missing From One or Both
Missing from AI Overviews
Focus on topical authority first. Are you publishing consistently on your core topic? Do other credible sites mention or link to you? Are your most important pages ranking on page one for their target keywords? If the answer to any of these is no, AI Overviews citation will remain elusive. These are not quick fixes. They take months. But they are the right problems to solve.
Also audit your content for clarity. AI Overviews prefers short, confident answers near the top of a page. If the answer to the query appears in paragraph twelve, restructure so it appears in paragraph one or two, with the detail following underneath.
Missing from AI Mode
This is where structured data does a lot of heavy lifting. Add or audit your schema markup across all key page types. Product pages need Product schema with accurate pricing and availability. Blog posts covering questions benefit from FAQPage schema. How-to content benefits from HowTo schema. About pages and author profiles should carry Person and Organisation schema to help the model understand who is behind the content.
Beyond schema, look at your headings. Do they use the natural language of questions your customers actually ask? Heading tags are one of the primary signals AI Mode uses to locate relevant sections within a long page. If your H2s are clever but vague, rewrite them to be specific and question-shaped.
For a full breakdown of how different content and schema strategies stack up against each other, our post on more content or better schema is worth reading before you decide where to spend your time.
The Bigger Picture
Google is running two parallel citation systems with different incentive structures. Treating them as one thing and optimising for both with the same strategy is a mistake most brands are currently making. AI Overviews rewards what traditional SEO has always rewarded: authority, ranking, and topical consistency. AI Mode rewards something newer: structural clarity, specificity, and the ability of a machine to extract a confident answer from your page in under a second.
The brands that will do well across both are the ones that take both seriously. That means investing in domain authority the long way, through good content and genuine mentions, while also making every page on the site as machine-readable as possible through proper schema implementation and precise, question-shaped content.
If you are not sure where your site sits right now, booking a walkthrough call is a fast way to get a clear picture. It takes thirty minutes and you leave with a list of specific things to fix, not vague recommendations about "improving your content strategy".
Frequently Asked Questions
Can a page appear in both AI Overviews and AI Mode?
Yes, and ideally it should. A page with strong authority and ranking position, combined with clear structure and schema markup, can be cited in both environments. They are not mutually exclusive, but they do require different optimisation priorities.
Does schema markup directly affect AI Overview citations?
Schema markup has a more direct and measurable impact on AI Mode citations than on AI Overviews. That said, certain schema types like FAQPage and HowTo do influence how Google's systems parse and present your content, which can indirectly improve your chances in AI Overviews too.
Why does my competitor appear in AI Mode but not rank well traditionally?
AI Mode is less tightly bound to traditional ranking signals than AI Overviews. If your competitor's pages are highly specific, well-structured, and use schema markup effectively, they can be cited in AI Mode conversations even without top-ten rankings. This is one of the reasons structured data matters so much in the current environment.
How quickly can citation performance change after optimisation?
AI Mode citation can shift relatively quickly, sometimes within weeks of implementing better schema and restructuring content. AI Overview citation tends to track more closely with traditional SEO timelines, meaning months rather than weeks, since it depends on ranking improvements and authority growth.