How to get recommended by Gemini

Gemini SEO means making your site clear, crawlable and useful when Gemini builds an answer. That matters this year if Gemini is part of your acquisition plan, because grounded Gemini answers can draw on content from the Google Search index at prompt time.

How is Gemini SEO different for you?

Gemini uses information Google has available, then turns that information into an answer to a prompt. That makes its route closer to classic search than the routes we see in other engines. In how CiteGraph measures, we recorded Gemini citing an average of 4 sources per answer, with the scanned product named in 6.8% of answers. Perplexity cited 5.3 sources per answer and named the scanned product in 8.2%, while the ChatGPT family cited 3.4 sources and named it in 4.4%.

Those percentages are not rankings. They show that an answer can cite several pages without naming the product those pages describe. Gemini recommendations therefore have two gates: your site must be available to Google’s systems, and the wider web must give Gemini usable evidence for including you.

The route starts with Google’s index.

Classic search work still matters, but the target page must also answer the buyer’s question without making the reader assemble the facts. A homepage that says little beyond a slogan may be indexed and still be a poor source for a recommendation. A comparison page, pricing page or clear product page gives Gemini material it can reuse.

Google’s documentation on common crawlers says that common crawlers find information for building Google’s search indexes. It also says that preferences addressed to the Googlebot user agent affect Google Search and its features. Read that in the right order: Googlebot controls the route into search, then Gemini has a chance to use what search knows.

How do you make Googlebot your first priority?

Start with access. Check robots.txt, server responses, sitemaps, canonical tags and rendered content. The free AI readiness check found that the AI crawler check passed on 116 of 121 audited sites, while the check for content present without JavaScript passed on 92 of 121. Those are not Gemini results, but they show where a site can lose usable content before a model has anything to recommend.

Googlebot is the important user agent here. Google says common crawlers always obey robots.txt rules, and describes Googlebot Smartphone and Googlebot Desktop as crawlers used for Google’s search systems. If you block those crawlers, or serve them a page without the important facts, you are making the index poorer before Gemini enters the picture.

Do not spend your first technical hour on Google Extended.

Google Extended does not have a separate HTTP user agent. Google’s documentation says it is a standalone product token that publishers can use to manage whether content may be used for training future Gemini models and for grounding in Gemini Apps. The same page says Google Extended does not affect inclusion in Google Search and is not a Google Search ranking signal. That makes it separate from the basic Gemini visibility problem described here. Blocking it does not block Googlebot, and allowing it does not guarantee a Gemini recommendation.

The settings control different things. Googlebot gets the page into Google’s search systems. Google Extended governs a separate use of content. For Gemini grounded in Google’s index, investigate Googlebot first.

Use Search Console and your server logs to check whether Google can fetch the important URLs, then inspect the live page as a reader would. The free AI visibility checker can help you look at the answer side as well, but neither tool turns a crawl pass into a recommendation.

Access is a gate, not a result.

Which pages should you build for Gemini?

Once Google can reach the site, make the facts easy to extract. State what the product is, who it is for, what it costs, and how it compares with the alternatives. Put those answers in visible text rather than relying on interface elements or a script that may fail to render. The readiness sample found that the plain statement check passed on 20 of 121 sites, and the comparison or alternatives check passed on 54 of 121. Those were among the checks that failed or warned most often.

The citation pattern points to a second job: earn pages outside your own site. In our recorded citations, listicles and roundups accounted for 67.5%, competitor product pages for 24.4%, community threads for 3.4%, and the scanned product’s own site for 1.5%. Together, listicles, roundups and competitor product pages accounted for 91.9% of citations. The weekly research on who owns AI answers shows the broader page mix.

The surrounding web matters more than most product teams expect.

Build pages that answer the questions buyers actually ask. A product page should explain the product. A pricing page should make the price visible or link clearly to pricing. An alternatives page should explain where the product fits and where it does not. A comparison page should use specific facts rather than vague claims. These pages help a model write an answer, and they give other publishers something concrete to cite.

The readiness data gives you a useful prioritisation rule. The price check passed on 90 of 121 sites, the title and description check on 92, and a clear H1 on 101. Those are easy wins, but they are not the whole Gemini plan. The harder gap is often the plain statement and the comparison page, where buyers need decision-making detail.

Here is what the aggregate numbers mean for the category as a whole, line by line.

1. The Gemini sample contains 1,784 answers.
2. Across those sampled answers, the scanned product was named 6.8% of the time.
3. The arithmetic is 1,784 × 0.068 = 121.312.
4. Rounded, that is about 121 answers naming the scanned product.
5. That figure describes the aggregate sample, not the expected result for one company or product.
6. Gemini cited 4 sources per answer on average, according to the same sample. That is an average across the answers, not a promise that one answer will cite 4 pages.
7. The citation mix shows where to inspect first: listicles and roundups accounted for 67.5% of recorded citations, and competitor product pages accounted for 24.4%.
8. If your product is missing from those answers, inspect both your own page and the independent pages that Gemini may use as evidence.

The calculation measures the sample’s scale. It does not predict what one rewrite will do.

There is a reasonable objection to this argument. Gemini is a model, so training data, brand familiarity or prompt wording may matter more than Google’s index in some answers. That can happen. But Google’s public documentation explicitly describes grounding as providing content from the Google Search index at prompt time to improve factuality and relevance. For the grounded Gemini use case, search access is still the first dependency.

What can’t you know yet about your Gemini visibility?

You cannot know from this sample whether Google has indexed every page you care about, whether a future Gemini answer will use a particular source, or whether a change to one page caused a recommendation. We measure samples, and we do not know what we have not measured. The Gemini sample contains 1,784 answers, which is useful evidence about the pattern, not a guarantee for your category.

That is the limit of the data.

You can still measure whether your site is accessible, whether relevant pages are clear, which answers name the product, and which sources Gemini cites. Compare those observations with how CiteGraph measures, then use the Lost answers report to look for missing prompt coverage or weak citation paths.

Report the evidence precisely. Do not say that a robots.txt pass means Gemini will recommend you, and do not treat a mention as proof that your page was cited. Record the prompt, the product mention, the cited URLs and the page types, then watch the same set over time.

The work is closer to SEO than to profile management. Your own site appeared in 1.5% of the citations recorded in the sample, so a strong product page still needs a surrounding web that can confirm what it does.

Return to the opening definition. Gemini SEO still means making your site clear, crawlable and useful. The 1,784-answer sample has changed the emphasis: you are not preparing for a mysterious Gemini channel, you are making the Google index and the pages around it do enough work that Gemini can use them when it answers.