How do you get Gemini and Google's AI answers to recommend your brand?

The first source is training. Google says large language models like Gemini train on content publicly available on the internet, and that content it crawls from websites may be used to train future generations of Gemini models unless the site refuses Google-Extended. What a model learned stops at its knowledge cutoff: Google's developer documentation describes grounding with Google Search as what lets Gemini cite sources beyond its knowledge cutoff.

The second is Google's own Search index. Google calls it grounding, which it describes as providing content from the Google Search index to the model at the moment it answers, and it names the Gemini app as one of the products that does it. So when the Gemini app grounds an answer, it reads from the same index as Google Search.

Google's AI answers in Search come from that index too. Google says AI Overviews and AI Mode are rooted in its core Search ranking and quality systems and retrieve pages from its Search index, and that both may use a “query fan-out” technique, issuing several related searches across subtopics to build one response. To appear as a supporting link, it says, a page must be indexed and eligible to be shown in Google Search with a snippet.

For a brand, that makes Google's index the lever that reaches furthest: a page in it can be drawn on by AI Overviews and AI Mode, and by the Gemini app unless the site refuses Google-Extended, with no wait for a new model.

Access is where Google publishes the most detail. It is split across four settings, and each reaches a different place:

• Googlebot. Google says robots.txt rules addressed to it affect all Google Search features, and that because AI is built into Search, those rules are the control over how a site is crawled for AI Overviews and AI Mode too.
• Google-Extended. Google publishes it as a control token rather than a crawler: it governs whether content Google has crawled may train Gemini models or ground Gemini's answers, and it does not affect Google Search.
• Search generative AI, in Search Console. A setting Google says decides whether a site's links and content can appear in AI Overviews, AI Mode and the generative AI features in Discover. Every site is included by default, and Google says the setting is not used as a ranking or inclusion signal for the rest of Search.
• Snippet controls. Google says nosnippet, data-nosnippet, max-snippet and noindex limit what Search shows from a page, and a page must be eligible for a snippet to be a supporting link in AI Overviews or AI Mode.

Google describes Google-Extended as one token covering both training and grounding, so a site that refuses it to stay out of Gemini's training also stops the Gemini app from using its pages to ground answers. Google says the token does not affect inclusion in Google Search, so those pages can still appear in AI Overviews and AI Mode. Google also puts allowing crawling in robots.txt, and at any CDN or hosting layer in front of the site, first among the SEO fundamentals it lists for AI features. Which crawlers to allow for the other assistants, and why a rule at your network edge can refuse one your file allows, is its own answer.

Then the pages themselves. Google says there are no additional requirements to appear in AI Overviews or AI Mode, and no special optimisation, beyond the SEO practices that already apply. Among the fundamentals it lists are important content available as text and structured data that matches the visible text on the page. Its newer guide adds that structured data “isn't required for generative AI search”, and says content people find unique, compelling and useful is likely, in the long run, to influence a site's presence there more than anything else it suggests. What an SEO team keeps doing is on our page about GEO and SEO.

Beyond that, the advice is ours rather than Google's. Buyers usually describe a situation rather than name a product: a small team, a budget, a named use. Answer that situation directly on the page, in a passage that still makes sense read on its own, since Google says AI Overviews and AI Mode can run separate searches for the parts of a question.

And the facts on them. Your own pages are among what Google indexes, so a wrong price, date or specification there can reach an answer. Where an answer and your page disagree, the fix can sit on either side. Which of these to start with depends on where your brand is weakest, and that order has its own answer.

You can see which of those pages Google's AI answers lean on in your own category. Ask a buying question in the Gemini app and open Sources where it appears: Google says the links may include public websites related to parts of the response, and that not every response has them. AI Overviews and AI Mode show their supporting links with the answer itself. Where the same domains keep coming back on questions a competitor wins and you do not, those domains are where the work goes. Why a competitor comes up first has its own answer.

Those pages sit at different distances from you, and the work changes with the distance:

• A listing you supply. A retailer page, a marketplace listing, a directory profile. You can influence it directly by correcting what you supply, though the page is not yours.
• An independent review or comparison. You can give the writer accurate facts, access to people who can answer questions and, where it fits, the product to try. What they conclude is theirs.
• A forum or community thread. You can answer the question openly, under the brand's own name. The opinion belongs to the people in the thread.

None of it moves on your schedule. A page about you changes when the people who publish it decide to change it.

Google's help for advertisers says ads can show above or below an AI Overview in every market where AI Overviews are available, and within the overview itself in English, on mobile and desktop, in a list of countries that includes the US. It says both the searcher's question and the content of the overview are considered when choosing those ads, and that an advertiser can neither target that placement nor opt out of it. When it launched ads inside AI Overviews, Google said they would keep its “Sponsored” label.

In AI Mode, Google is testing ads inside the answer itself. On 20 May 2026 it announced two new ad formats for AI Mode, built with Gemini, including one in which, when AI Mode gives a list of recommendations, “highly relevant, high-quality ads are eligible to appear on that list”. It says these formats will continue to be clearly labelled as sponsored.

So a sponsored entry can sit in the same list as the organic picks. Count the two apart: an entry labelled sponsored is an ad, however close it sits to the recommendations, and a report that counts ads and recommendations together overstates how often a brand is recommended.

Where the organic pages rank is Google's call. It says it does not accept payment to rank pages higher, and that its AI features in Search are rooted in its core ranking and quality systems.

The variation is large. In Delphi's study of answers collected up to 24 August 2026, an assistant asked the same question more than once inside one measurement named the identical set of companies 21.5% of the time. A before and after built from one answer each compares two draws, and it can show a gain or a loss that nothing caused. How much evidence a change needs has its own answer.

The free check by hand is written for ChatGPT, and the questions it sets out and what it says to write down for each answer carry over. Comparing a before with an after adds to what you record each time. Four things are specific to doing it with Google:

• Keep your own history out of the Gemini app. Google says the Gemini app can personalise its responses with memory of your past chats. Start a temporary chat, next to New chat: Google says a temporary chat does not give personalised responses.
• And out of AI Mode. Google says AI Mode draws on your past searches when history and personalised recommendations are on, and that you can opt out by changing your personalised recommendations setting.
• Ask from the same place, with the same model. Google says the Gemini app uses the general area from your IP address, or the home or work address in your Google account, by default, and that its Search rankings can take account of your location, language and device. The Gemini app also lets a signed-in user choose between models. Ask the before and the after from the same place, on the same device, with the same model selected.
• Allow for the clock. A rewritten page can only show up once Google has crawled and indexed it again, a third-party page once its publisher changes it, and a change in what the model learned only once a model with a later knowledge cutoff replaces it. Not every search shows an AI Overview: Google says one appears when its systems determine generative AI can be especially helpful.

Search Console shows part of it. Google says a site's appearances in AI Overviews and AI Mode are counted in its overall search traffic, in the Performance report under the Web search type, and since 31 August 2026 it has offered every site a generative AI performance report showing how often the site's URLs appeared in generative AI features in Search and Discover. Both count your pages, so an answer that names your brand without linking your site leaves nothing in either.

The study publishes that figure with the counts behind it and how each was made. A brand's standing in Gemini is one part of its AI visibility. Some of the work itself carries across and some does not. Accurate pages and third-party coverage are there for any assistant that reads the web. Crawler access is set operator by operator: most publish their own crawler, so a rule for Googlebot or Google-Extended says nothing about the rest.

The same question has its own answer for ChatGPT, Claude, Perplexity and Grok.

Delphi measures them side by side. Each time it measures a brand, it puts the same questions to ChatGPT, Gemini, Claude, Perplexity and Grok more than once and keeps every answer word for word. Its score, the Delphi Index, carries a margin set by how far the assistants disagreed with each other on your questions, and the dashboard and the history page do not report a change in the Index smaller than that margin as a change. Each assistant's share of answers naming you is shown beside the others' from one measurement to the next, so a change in Gemini alone is visible, though that figure carries no margin of its own.

Delphi asks each assistant through its API, the route developers use, rather than through the consumer app. Its questions to Gemini go out with Google Search switched on, and Google says the model then decides whether a search would improve the answer and, if so, runs one or more searches. So a page change Google has indexed can show in a Delphi reading of Gemini without waiting for a new model. Most of Delphi's questions reach ChatGPT without a live web search, so there the same change shows only once a newer model has learned it. Delphi does not read AI Overviews or AI Mode. Each paid level includes one or two measurements per brand a month. It is not a daily tracker, and an extra measurement costs $49 when a before and after needs one sooner.