AI search & AI transformation

From Traffic to Citations: How to Get Named in AI Answers

By Published Updated

Diagram: seven numbered circles in a loop around the words Revenue you can prove, step 6 of 7 highlighted. Each circle carries a letter for its job: A measurement, B AI answers, C AI-native team, D proof and positioning.

Step 6 of 7 in The Definitive Guide to Marketing in the Age of AI (in Seven Steps). Originally published December 16, 2024. Rewritten for the AI era October 26, 2026.

To get cited in AI answers, be where AI reads. That takes three channels working together. Earned sources: journalism, guest articles, podcasts, associations and partner sites. Owned pages built so a single passage can be quoted. Paid and in-person channels that still reach buyers. Measure success by whether you’re named and cited for your buyers’ questions, not by traffic alone.

A note on proof: Proof below predates generative AI; an AI-era result replaces it by mid-December.

Is website traffic still the right goal?

Not on its own. When Google shows an AI summary, people click a regular search result on 8% of visits, compared with 15% when there’s no summary. They click a link inside the summary on just 1% of visits, and they end their browsing session more often (26% vs. 16%) (Pew). Ahrefs found AI Overviews go with a 58% lower click-through rate for the top-ranked page (Ahrefs).

The buyer still learns something. They just learn it from the answer. If the answer names you, you’ve won the moment even without a visit.

Does being cited in an AI answer actually help?

Yes, though it doesn’t bring back the old traffic. Seer Interactive’s 2026 study of 53 brands found that being cited in an AI Overview delivered about 120% more organic clicks per impression than not being cited on the same results page. It still delivered 38% fewer than pages with no AI Overview at all (Seer Interactive). Seer is careful to call this correlation, not proof of cause.

So citations are the new rank. They’re the best position left on an AI results page.

Where do AI answers get their sources?

Mostly from people other than you. Muck Rack’s May 2026 study of more than 25 million links cited by ChatGPT, Claude and Gemini found:

  • 84% of citations came from earned media, a share that has held between 82% and 89% since July 2025.
  • 27% came from journalism alone.
  • 0.3% came from paid and advertorial content.
  • More than half of journalism citations came from articles published in the past 12 months.

Each engine behaves differently. ChatGPT cited sources in 96% of responses, averaging five. Gemini cited in 82%, averaging eight. Claude cited in 55%, averaging 13 (Muck Rack).

The takeaway is uncomfortable for content teams: your blog alone won’t get you named. It needs outside sources pointing to the same position and proof.

What is the citation flywheel?

It’s how owned, earned and in-person channels feed each other. Each turn makes the next one easier.

The citation flywheelA loop of four stages: Owned page, Earned placement, Third-party profiles, Sales and events, then back to the owned page. At the center: Named in AI answers.Owned pageEarned placementThird-party profilesSales & eventsNamed inAI answersThe citation flywheelA loop of four stages: Owned page, Earned placement, Third-party profiles, Sales and events, then back to the owned page. At the center: Named in AI answers.Named in AI answersOwned pageEarned placementThird-party profilesSales & eventsEach one feeds the next, then back to thetop
The citation flywheel: owned, earned and in-person channels feed each other.
  1. Owned page. One canonical page per buyer question, with an answer-first passage and dated proof (chapters 3 and 5).
  2. Earned placement. A guest article, podcast, association talk or press quote that repeats your position and links to the owned page.
  3. Third-party profiles. Speaker pages, partner listings and bios that describe you the same way.
  4. Sales and events. Reps and speakers use the same proof, which produces new stories, quotes and invitations.
  5. Back to owned. Each new result updates the owned page, so it stays recent.

Speaking is a good example of the earned turn. I’ve given more than 10 keynotes, at SMU, the University of Michigan and the University of Washington. A keynote reaches the room once. The host’s write-up, the recording page and the speaker bio keep describing you long after.

What are the steps to get cited in AI answers?

Generative engine optimization (GEO) sounds technical. Most of it is disciplined marketing. Here are the steps in order:

  1. Pick 5–10 buyer questions you want to own (from your question map).
  2. Run a baseline spot-check in ChatGPT, Gemini, Perplexity and Google AI Mode. Record who’s named and which sources are cited.
  3. Fix the owned page for each question: answer first, one idea per passage, named and dated proof, FAQ and schema.
  4. Target the sources the engines already cite. If one trade publication or association shows up again and again, that’s where to pitch.
  5. Earn one placement a month that carries your position and proof.
  6. Make your entity consistent across your site, LinkedIn, bios and directories.
  7. Re-run the spot-check monthly with the same questions, and log changes.

What still works from the old amplification playbook?

Organic and paid working together, events and sales as channels, and repetition. People still need to see an idea several times before it sticks. The difference is the destination. The original guide said every channel should point back to your website. Now every channel should point to the same position and proof, wherever the buyer meets it: in an AI answer, a podcast, a rep’s email or your site.

Checklist: from traffic to citations

  • 5–10 target questions, with a baseline spot-check saved
  • An answer-first owned page for each
  • A list of the outside sources AI already cites for those questions
  • One earned placement a month, carrying your position and proof
  • Consistent descriptions on every profile and bio
  • Paid judged on qualified conversations, not clicks

Where to go next

Citations and mentions are leading indicators. The real question is what grows revenue. Chapter 7 covers how to measure that when clicks disappear. If you want a baseline of where AI answers name you today, and what to fix first, ask about the measurement and AI-visibility audit: work with me.

You can’t buy your way into most AI answers. You have to be worth quoting, by people other than you.

Frequently asked questions

How do I get my company cited in AI answers?

Own a few buyer questions with answer-first pages, earn placements in the outside sources AI already cites for those questions, keep your company description consistent everywhere, and re-check monthly in four engines.

What is generative engine optimization (GEO)?

The practice of making your brand more likely to be named and cited in AI-generated answers. Most of it is clear positioning, checkable proof, answer-first pages and earned media.

Do AI Overviews reduce website traffic?

Studies say yes. Pew found people clicked a regular result on 8% of visits with an AI summary vs. 15% without, and Ahrefs measured a 58% lower click-through rate for the top-ranked page.

Does earned media matter for AI search?

Yes. Muck Rack’s May 2026 study found 84% of AI citations came from earned media and 27% from journalism, while paid and advertorial content was 0.3%.

Abstract illustration: on the left, a teal ring with a gap at the top marked by a dotted light-blue line, labeled Open loop; an arrow points right to a full ring of five arrowed steps around a light-blue center, labeled Closed loop.
Content that sells

Thought Leadership and Curriculum Fail the Same Way

In business, content is thought leadership. In education, it’s curriculum and companion materials. Both fail when the loop stays open, but only schools inherit a ready-made test.