What Your AI Citations Are Worth When Few People Click

Published: • 6 min read

A browser window showing a search for "best time to visit japan" with an AI Answer box at the top, a mouse cursor clicking beside a photo of Mount Fuji and cherry blossoms, and ordinary search results below.

ChatGPT

AI citations are the links an AI answer attaches to the pages that back up its claims, and their value depends on what the answer says about your company. A page can earn the citation while the AI recommends a competitor, so count recommendations on your buying questions before you count citations.

Key Takeaways

  • AI citations record which page backed a claim in an AI answer. Whether the AI recommended your company is a separate event.
  • Brands earn more organic clicks on Google searches where an AI Overview cites them, though the study shows correlation only.
  • Citations are easier to win and measure than recommendations, and a page change that lifts AI traffic can lose Google traffic.
  • Reality check: tracking recommendations means asking 10 buying questions by hand each week, and the answers can change between runs.

Executives keep asking whether an AI citation is worth anything if few people click it. A sharper question comes first: what did the answer say about your company? A citation can sit under an answer that recommends someone else.

What does an AI citation record?

Ask a question in Google, where an AI-written summary now sits above the results (called AI Overviews), or in ChatGPT, and the answer comes with links to the pages it drew on. Those links are the citations. Google’s documentation for site owners says AI Overviews respond with links to supporting websites (1. Google Search Central, n.d.). A citation records that a page backed up a claim in the answer. Nothing in that definition stops an answer from citing your page and recommending a competitor.

Which company the answer then recommends is a separate event, and the citation count says nothing about it. Cyrus Shepard, who runs the SEO research company Zyppy, made the same split on a recent webinar hosted by SearchPilot, a company that runs controlled tests on websites. In his own smaller experiments, citations were easy to win, at least for a while: follow a formula and put the key facts near the top of the page. He said recommendations are harder to earn and citations are much easier to measure, and that he worries the industry is concentrating on the wrong one (2. SearchPilot, 2026).

Do AI citations send any clicks?

A few. Pew Research Center, a nonpartisan research organization, tracked the browsing of 900 U.S. adults in March 2025. When a Google results page showed an AI summary, people clicked a link inside the summary in 1 percent of visits (3. Pew Research Center, 2025). That’s Google only, and more than a year old.

Seer Interactive, a search marketing agency that sells AI search services, tracked 53 brands through February 2026. Take informational searches. When an AI Overview cited a brand, it earned 120 percent more clicks on its regular results (organic clicks) each time they appeared than when the Overview left it out (4. Seer Interactive, 2026). Seer says it can’t claim cause and effect, because stronger brands are also more likely to get cited. Even when cited, a brand earns fewer clicks than it would on searches where Google shows no overview.

Seer adds that a link in an overview can lead people to run a fresh Google search, so part of a citation’s effect lands off your site (4. Seer Interactive, 2026). Shepard compares the Seer result to buying a search ad and then ranking organically for the same phrase, which lifts organic clicks. He says he hasn’t seen the study replicated (2. SearchPilot, 2026).

The data supports few clicks on Google, a possible lift for cited brands, and an effect you can only partly see. Saying “zero clicks” rounds a small number down to nothing.

What do buyers do with an AI recommendation?

G2, the software review site, surveyed 1,076 B2B software buyers in March 2026. Of those buyers, 69 percent said they chose a different vendor than they had planned because that vendor appeared in a chatbot’s recommendation (5. G2, 2026). G2 also argues that a background citation in a long research report won’t win a deal, while being one of the brands recommended at the end of it can.

Two cautions apply. G2 sells marketing services to the software vendors it lists, and the figure is buyers describing their own behavior. Still, it points at the event your citation count skips: the moment the answer names a vendor.

Can optimizing for AI answers hurt your Google traffic?

It can, and one tested case shows how. Omio, a travel booking platform, ran two page changes through SearchPilot, both built to make pages easier for AI assistants to use. The tests measured visits from those assistants alongside Google traffic. The first change added clear brand selling points. It lifted traffic from AI assistants, and SearchPilot found no evidence of harm to Google traffic. The second added bulleted comparison data above an existing summary table. It also lifted AI-assistant traffic, and SearchPilot estimated it would likely cut Google organic sessions. Google is Omio’s largest traffic source, so Omio held that change back (6. SearchPilot, 2026).

SearchPilot sells this testing, and the page is its own customer story. Omio counted the second test as positive for AI traffic, and a team watching only AI results would have shipped it. Shepard reported the same pattern in his smaller experiments: wins in AI answers and losses in Google (2. SearchPilot, 2026).

What should you count instead of AI citations?

Count recommendations on the questions your buyers ask. Write 10 questions a buyer in your market would put to an AI before contacting a vendor. A CMO replacing a marketing automation platform might ask which platforms suit a mid-market B2B company and what to check before signing. Ask each one weekly in ChatGPT, Perplexity, and Google’s AI Mode. Log one of 3 results: recommended, cited only, or absent.

Repeat each question a few times. Ahrefs, an SEO software company that tracks AI Overview citations at scale, says the overviews can return different answers to the same question (they’re probabilistic), so one run proves little (7. Ahrefs, 2026).

Then add two signals you already have. Check how many people search for your company name (branded search volume) in Google Search Console. An overview link can lead people to a fresh Google search, and that search may be for your name. And put “how did you hear about us” on your demo form.

The method measures the event buyers act on. It’s manual, and it can’t prove an AI answer caused a deal. Bring the 10 questions to your next agency or vendor review and ask which ones recommend you today. Then ask what the team will change to move the rest, and what that change could cost you in Google.

About the Author

Gene De Libero, Independent Martech Advisor, Digital Mindshare LLC

Gene De Libero has spent more than thirty years in marketing technology — as buyer, seller, builder, and advisor. He is the creator of the Marketing Technology Transformation® framework, sponsor of How Marketing Technology Works®, and Independent Martech Advisor at Digital Mindshare LLC, a New York consultancy serving CMOs whose stacks have stopped paying for themselves. He believes most martech investments fail not because the technology is wrong, but because the organization was never built to use it. He fixes that.

Frequently Asked Questions

Do AI citations send traffic to your website?

Some, but few. Pew Research Center found people clicked a link inside a Google AI summary in 1 percent of visits. Seer Interactive found brands earned more organic clicks on searches where an AI Overview cited them, though that study shows correlation only. Clicks exist, and they’re rare.

Are AI citations the same as AI recommendations?

They’re separate events. A citation is a link that backs up a claim in the answer. A recommendation is the AI naming your company as an option for the buyer. A page can earn the first while a competitor gets the second, so tracking only citations can overstate your standing.

Can optimizing for AI answers hurt your Google traffic?

It can. In SearchPilot’s test for Omio, a travel booking platform, a page change that raised traffic from AI assistants was projected to cut Google organic sessions by 6.5 percent, so Omio held it back. Measure both channels in the same test before you roll a change out.

How do you check whether AI recommends your company?

Write 10 questions a buyer would ask before contacting a vendor. Ask each one weekly in ChatGPT, Perplexity, and Google’s AI Mode, repeating it a few times because answers vary. Log each result as recommended, cited only, or absent. Add branded search volume from Search Console as a second signal.

Do buyers use AI to choose software vendors?

Many do. In G2’s March 2026 survey of 1,076 B2B software buyers, 51 percent said they start research with an AI chatbot more often than with Google. The figure is self-reported, and G2 sells marketing services to the software vendors it lists.
References
  1. Google Search Central. (n.d.). AI features and your website. Google. https://developers.google.com/search/docs/appearance/ai-features
  2. SearchPilot. (2026, October 2). Winning AI citations isn’t the same as winning in search | With Cyrus Shepard [Webinar]. https://www.searchpilot.com/resources/blog/winning-ai-citations-isnt-the-same-as-winning-in-search-with-cyrus-shepard
  3. Pew Research Center. (2025, July 22). Do people click on links in Google AI summaries? https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results
  4. Seer Interactive. (2026, April 24). AIO impact on Google CTR: 2026 update. https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update
  5. G2. (2026, April 15). In the Answer Economy, don’t win the click: Win the answer. https://company.g2.com/news/g2-research-the-answer-economy
  6. SearchPilot. (2026). GEO A/B testing: SEO and GEO are not the same [Customer success story]. https://www.searchpilot.com/geo-a/b-testing-seo-and-geo-are-not-the-same
  7. Ahrefs. (2026, March 2). Update: 38% of AI Overview citations pull from the top 10. https://ahrefs.com/blog/ai-overview-citations-top-10/