How do brands get cited in Google’s AI Overviews?

What changed in Google’s AI answers this month?

On 15 October 2026, Google announced five changes to how AI Mode and AI Overviews show links to websites. Links now appear inline inside the answer text, higher and more often. Google is testing a box at the top of a response that highlights one or two official sites for a task. News carousels that surface a user’s Preferred Sources are expanding to more locations and languages. End-of-answer links get image thumbnails and a more clickable design. And clickable publisher and creator names started appearing on AI Mode video results in the week of 12 October, as reported by Search Engine Roundtable.

Google also said use of Preferred Sources has grown more than sevenfold since the start of 2026, per PPC Land. Google released no click or traffic data alongside the announcement, and told Search Engine Roundtable it will “introduce additional metrics over time.”

We read this as a direction of travel rather than a fix. The answer box is not going away. What is changing is that a citation inside it now carries a visible, clickable link in more places. That makes being named in the answer the goal, not a consolation prize.

What happens to clicks when an AI Overview appears?

Four independent studies now point the same way. The sharpest evidence comes from a randomised field experiment by Saharsh Agarwal of the Indian School of Business and Ananya Sen of Carnegie Mellon, covering 1,065 US participants and 68,089 searches.

Two details matter for planning. First, the losses sit in informational queries: overviews were triggered or intended for 53% of informational searches, against 15% navigational and 6% transactional. Second, Pew found that searches phrased as questions produced a summary 60% of the time. The queries most likely to introduce a brand to a new buyer are the ones most likely to be answered without a click.

How does Google choose which sources to cite?

The citation pattern is public, even if the ranking system is not. Pew found that Wikipedia, YouTube and Reddit were the three most cited sources in AI summaries, making up 15% of all sources. Government sites accounted for 6% of AI summary sources against 2% of standard results. News sites were 5% of both.

A larger sample tells the same story. Peec AI analysed 30 million cited sources across ChatGPT, Google AI Mode, Gemini, Perplexity and AI Overviews, reported by Search Engine Land in March 2026. Reddit was the most cited domain on all five platforms, followed by YouTube and LinkedIn, with Wikipedia and Forbes also in the top five. Review platforms such as Yelp and G2 appeared often on recommendation queries.

Two things follow. Brand-owned pages are not the main raw material for these answers. Conversation is. And YouTube citations draw on transcripts and descriptions, which means a creator video can carry a brand into an AI answer that the brand’s own site never reaches.

This is the part most AIO advice misses, and it is where our two practices meet. Pages and schema give the model something accurate to quote. Creators, forum threads, reviews and vernacular video give it something to quote from. We track both sides through Brand Chatter Intelligence, because an AI answer that recommends a competitor is usually reflecting what people said, not what the site said.

Is this only a US problem?

No. AI Mode is now available in more than 200 countries and territories, after an expansion in October 2025 that added over 35 languages and more than 40 new markets, including Indonesia, Japan, South Korea and Australia, as documented by PPC Land. India got AI Mode in June 2025, ahead of the United Kingdom in July. Hindi, Indonesian, Japanese, Korean and Portuguese support arrived in September 2025.

That sequencing matters for anyone selling across our markets. The US and UK have the deepest published research, so that is where the measured click loss comes from. But India and Indonesia were early, not late, and both are now served in their own languages. A Hindi or Bahasa Indonesia question gets an AI answer built from Hindi and Bahasa sources, which are thinner and easier to influence than the English corpus.

For the UAE and the wider Gulf, the same logic applies in Arabic and in expat English. Google has not published per-market click data for the region, so we do not quote any. What we do instead is measure it directly for clients: a tracked set of buying questions, run weekly in each market and language, with the cited domains logged each time.

In India we have one further advantage worth naming plainly. Our network reaches 350,000+ regional creators across 22+ languages and 650+ cities. When the cited corpus for a Tamil or Marathi question is small, a credible regional creator can move it.

How do you earn AI Overview citations, step by step?

This is the sequence we run. It takes about a quarter to show movement.

1. Build a question list, not a keyword list. Write down 50 to 100 questions a buyer actually types, in every language you sell in. Pew’s finding that question-phrased searches trigger a summary 60% of the time tells you which phrasing to prioritise.
2. Check what the answer says today. Run each question in AI Mode and in AI Overviews from each market. Log the answer, the cited domains and whether you appear. This is your baseline, and nothing else in the list means much without it.
3. Separate the queries you can win. Informational questions are where overviews dominate and where citations are winnable. Navigational and transactional queries barely changed in the randomised study, so keep conventional SEO working there.
4. Rewrite your pages to be quotable. Put a direct two-to-three sentence answer under each question heading, before any context. Give specific numbers, dates and named examples. Models quote what is unambiguous.
5. Add the structure machines read. FAQPage and Article schema, clean headings, a visible publish and update date, a named author with real credentials, and internal links between related answers.
6. Get into the sources that get cited. Earn genuine presence on Reddit, YouTube, LinkedIn and the review platforms in your category. For YouTube, publish full transcripts and detailed descriptions, since that is the text being read.
7. Brief creators for answers, not just reach. We give creators the question, not only the product. A video titled around the buyer’s actual question earns citations that a brand film never will.
8. Re-measure monthly and keep a citation log. Track share of tracked questions where you are cited, by market and language. Treat that share as the primary metric, with referral traffic as secondary.

One caution. Google still gives publishers no click data for AI features, so attribution stays incomplete. Anyone promising clean AI Overview ROI reporting today is overselling.

Frequently asked questions

Shamik TalukderCo-founder & CEO, BeTagged

← More from Insights