How to Rank in Google AI Overviews

To rank in Google AI Overviews, make sure Googlebot can crawl and index your page, and answer the searcher's exact question in plain text that Google can show as a snippet. Google says there are no extra requirements, special markup or AI text files needed, because AI Overviews link to pages from its normal search index.

An AI Overview is the AI-written summary Google shows above the regular results, with links to the pages it drew from. Being one of those links is what "ranking" in an AI Overview means. We checked how this plays out in India: on 15 September 2026 we pulled 60 Google results pages for 10 shopping questions, each asked in English and in Hinglish. Hinglish searches showed an AI Overview in 28 of 30 captures. English searches showed one in 18 of 30.

Hinglish here means Hindi and English mixed and typed in English letters, like "konsa face wash kharidna chahiye". The rest of this guide covers what Google says, what our data shows about which pages get linked, and 8 steps you can take this week.

What decides which pages AI Overviews cite?

Google's AI features and your website page (last updated 10 December 2025) sets one condition. To be shown as a supporting link, a page "must be indexed and eligible to be shown in Google Search with a snippet". A snippet is the short text preview under a result. The same page says there is no special schema.org structured data to add and no new machine-readable files to create.

Everything else on Google's list is normal SEO: allow crawling in robots.txt, link to important pages internally, keep important content in text, use good images and video, make structured data match the visible page, and keep Merchant Center and Business Profile details current. Robots.txt is the file on your site that tells crawlers which pages they may fetch.

Google also says AI Overviews may use "query fan-out". That means Google runs several related searches across subtopics behind the scenes and builds the answer from all of them. So a page can be linked even when it does not rank for the exact question typed.

Our pull shows both effects. Across the 41 Overviews whose links loaded, we counted 274 distinct web pages linked:

• 95 of 274 links (34.7%) were pages that also appeared among the 7 to 9 regular results on page one of the same search.
• 40 of 41 Overviews linked at least one of those page-one results.
• 16 of 41 linked the number 1 regular result.
• In English, 26.7% of links were page-one results. In Hinglish, 40.9% were.

Basis: Google results pages pulled from India on 15 September 2026, desktop and mobile, 10 shopping questions in English and Hinglish, 3 captures each. Page one helps: 40 of 41 Overviews used at least one page-one result. Still, 179 of the 274 links went to pages outside page one, which fits Google's description of fan-out.

How often do AI Overviews show for Indian shopping questions?

We used the 10 buying questions from our Hinglish vs English AI shopping study: 5 skincare and 5 fashion, none naming a brand, each with an English and a Hinglish version of the same meaning. Each version was searched 3 times: once on desktop, once on mobile, and once more on desktop. The table shows in how many of the 3 captures an AI Overview appeared.

Three patterns stand out:

1. Fashion questions in English rarely got an Overview. English skincare questions showed one in 12 of 15 captures. English fashion questions showed one in 6 of 15. In Hinglish, both categories showed one in 14 of 15.
2. English pages without an Overview showed products instead. All 12 English results pages with no Overview carried Google's product carousel, the row of product cards with prices. None of the 30 Hinglish pages showed that carousel.
3. The same question could flip between captures. 4 English questions got an Overview in only 1 of 3 captures, all taken within about 20 minutes. One check tells you little.

Which sites did AI Overviews cite?

We counted how many of the 41 Overviews with loaded links cited each site at least once. 68 different sites were cited in total. The table shows the 10 cited most often.

Grouped, 34 of 41 Overviews cited at least one of Amazon.in, Flipkart, Nykaa or Myntra. 24 of 41 cited YouTube or Instagram. 14 of 41 cited Reddit or Quora.

75 of the 274 linked pages (27%) sat under a blog path, and 76 (28%) had "best" or "top" in the web address. Many were buying guides, with web addresses like "10-affordable-face-serums-under-500-in-india" and "top-jeans-brands-in-india-men". Brand sites were cited too. Skinaa, Dot & Key and Plum were each cited in 6 English Overviews. Skinaa was linked through buying guides on its own blog. Dot & Key and Plum were linked mostly through category pages named after the buying need, such as Plum's "face washes for oily skin" collection.

Two more things showed up only because we checked both languages:

• Hinglish Overviews were mostly written in Hindi script. 21 of 25 Hinglish Overviews with loaded text were written mainly in Devanagari, the script Hindi is usually printed in. The person searching had typed in English letters.
• Google translated English pages for Hinglish readers. 19 of 25 Hinglish Overviews linked at least one English page through Google Translate, 49 of their 154 links in total. Your English buying guide can be cited in a Hindi answer.

Google's own shopping listings also appeared inside the answers. 17 of 18 English Overviews and 16 of 25 Hinglish Overviews included product cards that open Google's shopping product viewer.

8 steps to get your pages into AI Overviews

1. Let Googlebot crawl the page. AI Overviews link to pages in Google Search's index, and Googlebot is the crawler that fills it. Check robots.txt, firewall and CDN rules, then run the page through URL Inspection in Search Console. Our AI crawler guide explains which bots feed search and which collect training data.
2. Decide on Google-Extended separately. Google's common crawlers documentation says Google-Extended controls use of your content for training Gemini models and does not affect inclusion in Google Search. Blocking it will not take you out of AI Overviews.
3. Keep snippets switched on. Google's eligibility rule needs a page that can be shown with a snippet. A nosnippet tag, or a very short max-snippet value, limits what Search can show from that page. Use them only on pages you want kept out.
4. Rank on page one for the exact question. In our pull, 40 of 41 Overviews linked at least one page-one result. The classic work still counts: a clear title, a page that matches the intent, and internal links from related pages. Our GEO vs SEO comparison covers where the two overlap.
5. Answer in the first lines, in plain text. Put the direct answer in the first 2 sentences, then the detail. Google lists keeping important content in text form as a best practice. An answer that exists only inside an image or a video is harder for Google to quote.
6. Publish buying guides for real buying questions. 28% of linked pages in our pull had "best" or "top" in the web address. Name the products, give prices, say who each one suits, and date the page. If you sell the product, your own pages can be cited too: Dr. Sheth's guide to sunscreens with no white cast and Plum's oily-skin face wash collection both were.
7. Be present on the sites Overviews already cite. Marketplaces appeared in 34 of 41 Overviews, YouTube or Instagram in 24, Reddit or Quora in 14. Keep your Amazon.in, Flipkart, Nykaa and Myntra listings complete, keep Merchant Center data current, post useful videos on YouTube, and answer real buyer questions on Reddit. Citation gap analysis shows the sites that recommend your competitors and never mention you.
8. Check Hinglish searches on their own. Hinglish showed an Overview in 28 of 30 captures against 18 of 30 for English. 40.9% of Hinglish Overview links were page-one results, against 26.7% for English. If your buyers type in Hinglish, test those searches separately and keep your English guides easy to translate: short sentences, product names in text, prices as numbers.

How to check if an AI Overview shows for your queries

You can check a handful of searches by hand:

1. Open a private browser window so your history does not shape the results.
2. Go to google.co.in and type the question exactly as a buyer would, in English and in Hinglish.
3. Note whether an AI Overview appears. If it does, expand it and write down every linked site.
4. Repeat on your phone and again later in the day. In our pull, 5 of 10 English questions changed between 3 captures.

Search Console helps only in aggregate. Google says traffic from AI features is included in the overall Performance report, under the Web search type.

Manual checks get slow past a few questions. Depra's Google AI Overview tracker checks your buying questions daily on a real Google results page from inside your country. It records whether an Overview appeared at all, whether it named your brand, and which pages it linked, in English and Hinglish. How often Overviews appear is kept apart from whether you are in them, so a week with fewer Overviews does not look like a drop in your visibility. The AI search glossary explains the terms used in AI visibility reports.

What are the limits of this data?

• One day, one short window. All 60 pages were captured on 15 September 2026 between 13:50 and 14:11 IST. Google changes when it shows Overviews, so treat this as a snapshot.
• 10 questions in 2 categories. Skincare and fashion buying questions with no brand names. Informational or local searches may behave differently.
• India, logged out, country level. Searches were run from India with Google's interface language set to English, and no signed-in account. A signed-in user in a specific city may see something else.
• 3 captures per version. Desktop, mobile and a desktop repeat. That shows variation between captures and is too small to estimate a precise rate.
• Capture gaps in the last pass. In the desktop repeat, 3 Hinglish Overviews appeared without their text or links loading, and 2 Hinglish pages showed no Overview. These came during a stretch of retrieval errors, so the Hinglish count may be low. The 3 empty Overviews count as shown in the table and are left out of the citation counts.
• One page counted once. Google adds a tracking tag (srsltid) to the end of many shopping page addresses, and the tag changes from link to link. We removed it before matching and counting, so the same page is counted once.
• Page one means 7 to 9 regular results. The rest of each page was taken up by the Overview, videos, product rows and "People also ask".
• Links show what Google cited. They do not show why. Our counts describe which sites and page types were linked. They do not prove that a given change will get you linked.
• Funding interest. Depra ran and paid for this pull and sells AI Overview tracking.

To see whether AI Overviews show for your own buyers' questions, and which sites they cite in English and Hinglish, start 7 days free on any plan, no card. Your first scan starts at signup.