Think of an AI mention or citation as an impression. It puts your brand in front of a buyer right when they decide, and it rarely brings you a click. In Pew's data only 1% of Google visits with an AI summary included a click on a cited link1. So judge citations by reach and recall, never by CTR.
Out of 100 Google visits, how many reach a website?
What's the difference between a mention, a citation and a click?
People lump these three together all the time. They do very different jobs.
| What happens | What it's closest to | |
|---|---|---|
| Mention | The answer names your brand, with no link | An ad impression with no click button |
| Citation | The answer links to your page as a source | An impression with a small, rarely used link |
| Click | A person leaves the answer and lands on your site | A visit, like any other referral |
A mention or a citation is exposure. A click is a visit. Most reporting treats the first two as if they should produce the third.
Why do most people expect citations to bring clicks?
Twenty years of SEO taught you one simple chain. You rank, people see you, people click. Rank one got the most clicks, so visibility and traffic moved together.
AI answers broke that chain. The answer does the reading for the person. When the question is answered on the page, there's nothing left to click for.
So here's what usually happens. Teams add "AI citations" to the dashboard, watch referral traffic stay flat, and decide AI search doesn't work for them. That's the mistake I see most often. They're judging an impression channel by click-through rate.
How often do people click a link in an AI answer?
Honestly, almost never. Pew tracked real browsing in March 2025. With an AI summary on the page, people clicked a normal result in 8% of visits. Without one, they clicked in 15%. Links inside the summary got a click in 1% of visits1.
People also just stopped searching. Visits with an AI summary ended the browsing session 26% of the time, against 16% without one1.
Ahrefs measured the same effect from the publisher's side. By December 2025, an AI Overview cut the click-through rate of the top-ranking page by 58%, compared with similar keywords without one2. In April 2025 the same study found 34.5%. And the gap keeps growing.
Being cited doubles your clicks against rivals. It still loses to no AI answer at all.
Does being cited still help?
It does, compared with the brand sitting next to you. Seer Interactive found cited brands earned 2.07% organic CTR on informational queries in 2025. Uncited brands on the same kind of page earned 0.94%3.
That's 120% more clicks per impression. It's also 38% below pages with no AI Overview at all3. A citation wins the fight on the page. It doesn't bring back the old traffic.
Why is a citation closer to an impression than a click?
Here's the part I find most interesting. Look at what happened to cited pages in Seer's data between September and October 2025. Impressions more than doubled, from 15.8 million to 33.1 million. Clicks stayed flat, at about 400,0003.
CTR halved, from 2.52% to 1.21%. A team watching CTR would call that a disaster. A team watching reach would call it a great month. The brands were shown twice as often to the people asking.
Reach doubled. Clicks didn't move.
Do mentions without a link count?
They count, and they count more than links do. Ahrefs studied 75,000 brands and what predicts how often Google's AI Overviews name them. Branded web mentions had a correlation of 0.664. Backlinks had 0.2184.
And the effect is really lopsided. Brands in the top quarter for web mentions got a median of 169 AI Overview mentions. The next quarter got 14. About 26% of brands got none4.
A mention on someone else's page sends nobody to your site. It still teaches the model who you are. That's an impression in the oldest sense, a mark left in memory.
What predicts being named in Google's AI Overviews?
Why don't AI tools send traffic back?
Because reading your page and sending someone to it are two separate jobs. Cloudflare compares how many pages each platform's bots fetch with how many visitors that platform refers back.
In July 2025, Google fetched about 5 pages per referred visit. Perplexity fetched about 195, OpenAI about 1,091 and Anthropic about 38,0665. The ratios swing month to month, so read them as orders of magnitude.
The totals stay small. SE Ranking found AI platforms sent 0.32% of all website traffic in 2026, up from 0.24% in 20256. It's growing fast, just from a tiny base.
Pages read for every visitor sent back
What should you measure instead of clicks?
Treat AI visibility the way brand marketers treat TV or billboards. Count reach, then look for the lift downstream. Here's the stack I'd set up.
Five numbers to track, from answer to deal
Share of answers. Write 30 to 50 prompts your buyers actually ask. Run them every month in ChatGPT, Gemini, Perplexity, Claude and Google AI Mode. Record whether you're named and who's named instead. That's AI answer monitoring.
Citation share. Of the answers that name you, count how many link to your page. This shows which of your pages the models trust as a source.
Branded search. People who meet your name in an answer often search for it later. Watch branded queries in Search Console against your share of answers.
Direct and AI visits. Much AI traffic arrives with no referrer and lands in "direct". Track direct sessions to deep pages, plus the AI referrals you can see.
Self-reported source. Ask "How did you hear about us?" when people book a call. Buyers will tell you "ChatGPT recommended you" when analytics can't.
And please stop reporting CTR on AI-cited queries as a success metric. A falling CTR with flat clicks can mean you're being shown more. Seer saw exactly that3.
How do you earn more mentions and citations?
Be the source. Publish numbers nobody else has. In the KDD 2024 GEO study, adding statistics and quotations raised a page's visibility in generative answers by up to 40%7.
Answer first. Put the answer in the opening lines. Models use information at the start and end of their context best, and lose track of it in the middle8.
Get talked about off your site. Web mentions predicted AI Overview mentions three times better than backlinks4. Reviews, podcasts, YouTube, Reddit threads and press all count.
Make your entity unambiguous. Use one brand name everywhere. Mark it up with Organization and Person schema, and link your profiles with sameAs. I build this layer as part of context systems.
Cover the sub-questions. Google says AI Mode and AI Overviews may run several related searches behind one question9. Each sub-question is another chance to be the cited page.
To see which of these tactics hold up in the research, read my check of GEO ranking factors. If you'd like the step-by-step version, have a look at how to get your site cited by ChatGPT. To start, map your site with the free AI website brief builder.
Questions people ask
Are AI citations worth chasing if they don't bring clicks?
Yes, they're worth it. They put your name in the answer while the buyer is deciding. Count them like impressions and look for the payoff in branded search, direct visits and what buyers tell you.
How do I track AI mentions?
Run a fixed set of buyer prompts in the main AI tools every month. Record whether you appear, whether you're linked and who appears instead. Paid tools automate this, and a spreadsheet works to start.
Does an AI Overview citation raise click-through rate?
Against uncited rivals, yes. Seer found 120% more organic clicks per impression for cited brands in 2025. That's still 38% below queries with no AI Overview, and Seer can't prove the citation is the cause.
Do backlinks still matter for AI search?
They still help rankings, and rankings feed some AI answers. In Ahrefs' study of 75,000 brands, web mentions correlated about three times more strongly with AI Overview mentions than backlinks did.
Why did my CTR fall while clicks stayed flat?
Your impressions grew faster than your clicks. Seer saw this on cited pages in October 2025, when impressions doubled and clicks held. Check clicks and impressions separately before you call it a problem.