Justin McKelvey
Ran snowboard shops. Taught myself to code. Now I build AI that does real work.
How to Rank in Google AI Overviews (2026): 29 Cited Sources, Zero From the Top 10
Quick Answer
To rank in Google AI Overviews, your page has to be indexed, allowed to show a snippet, and open with a specific answer to the exact question, with the number in the first lines. Google says there's nothing else to do. But ranking in the top 10 isn't what gets you picked: when I pulled three buyer questions on November 18, 2026, the AI Overviews cited 29 sources, and none of them were the pages in the top-10 blue links underneath. Google runs extra related searches behind every AI Overview, so it picks from a wider pool. Half of the cited pages on the consulting-cost question opened with a labeled short answer ("Quick Answer," "TL;DR") or with the question itself, followed by a number.
Tested November 18, 2026 · Author: Justin McKelvey, AI consultant and fractional CTO, Austin TX
What Google Says It Takes
Google's official answer is short: nothing special. Its Search Central page AI Features and Your Website (read November 18, 2026) says "there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." To be a supporting link, a page has to be indexed and eligible to show a snippet.
The same page has the sentence that matters more: AI Overviews and AI Mode "may use a 'query fan-out' technique," running several related searches across subtopics and showing "a wider and more diverse set of helpful links" than a classic search. In plain words, the AI box doesn't just summarize the top 10. It goes looking.
One-time · $29
Find out if ChatGPT recommends you, or your competitor.
- The 25-query worksheet that finds your money queries
- The 4-week tracking spreadsheet: query × engine × cited?
- The 15-action fix checklist, ranked by impact
Instant access. Your link lands on the next screen and in your email.
I Pulled Three Buyer Questions. Here's What Got Cited.
On November 18, 2026 I pulled the full Google results page (US, English) for three questions a buyer asks, with the AI Overview loaded, and compared the cited sources against the ten blue links underneath.
| Question | Sources cited in the AI Overview | Cited pages also in the top 10 links |
|---|---|---|
| How much does an AI receptionist cost? | 11: 4 YouTube videos, 2 Reddit threads, 5 web pages | 0 |
| How much does an AI consultant cost? | 10: 9 web pages, 1 Reddit thread | 0 |
| Claude team plan pricing | 8: 2 Anthropic help pages, 6 independent sites | 0 |
| Total | 29 | 0 |
The closest thing to an overlap: on the receptionist question, one company's blog posts held two of the blue links, while the AI Overview cited that same company's pricing page. Different page, different job.
This is one moment in time, three questions, one country. It isn't a study. But zero out of 29 is a strong enough signal that "rank first, then you'll be in the AI box" is the wrong mental model.
What the Cited Pages Had in Common
1. The answer comes first, often under a label. On the AI consultant cost question, 5 of the 10 cited pages opened with "Quick Answer," "TL;DR," "The Short Answer," or the question itself followed by a number. Alice Labs' pricing guide, for example, starts with "TL;DR Quick Answer" and the hourly ranges. Another opens: "Most independent AI consultants charge $150 to $350 an hour." On the Claude question, one cited page opened with "Short answer: Claude Free is $0."
2. Real numbers, stated as sentences. Every cited snippet I read had a price, a range or a count in its first lines. "On the Team plan, a Standard seat costs $25 a month and a Premium seat $125" is quotable. "Pricing varies by business" isn't.
3. The source of truth gets a seat. On the Claude question, two of the eight citations were Anthropic's own help center, including its Team plan setup article. If you sell the thing, your own help or pricing page is the most likely page to be cited, which is one more reason to publish your prices.
4. Video and forums count. On the receptionist question, more than half the sources were YouTube and Reddit. One of the Reddit threads was written by the co-founder of a receptionist company, answering honestly about his own pricing. That's a legitimate channel if you answer like a person, not an ad.
Want the checklist version of this? The free 12-point AI Discoverability Checklist: the self-audit for whether ChatGPT, Perplexity, and Google AI can find and cite your site.
Free. One email, no drip campaign ambush. One-click unsubscribe.
My Own Page Ranks and Still Isn't Cited
Here's the uncomfortable part. My Claude Team plan page averages position 6.9 for "claude team plan pricing" (Search Console, October 19 to November 16, 2026: 393 impressions, 1 click). It opens with a Quick Answer that states the per-seat price. The AI Overview cited eight other sources.
So ranking and an answer-first opening make you eligible. They don't guarantee the pick. Google pulled from pages that weren't even on page one. I'll keep testing what changes the outcome, and I'll add the results here as I get them.
A Checklist That Matches the Evidence
-
Make sure the page can show a snippet. Check for
nosnippet,data-nosnippetor a tightmax-snippeton your money pages. Google's snippet controls apply to AI features too. - Open with the answer, labeled. First lines: the question's answer with the number in it. A "Quick Answer" box is fine; the label isn't magic, the specific answer under it is.
- Write one page per question a buyer asks. Fan-out means Google searches the sub-questions too ("cost for a small business," "per-seat price," "is there a free plan"). A page that answers the sub-question exactly can be cited even if it never ranks for the main one.
- Publish your own pricing page. The vendor's own page was cited on two of the three questions.
- Consider a short video for cost and how-to questions. Four of 11 receptionist sources were YouTube.
- Measure, don't guess. Check your 10 most important questions monthly. How I track mentions across engines is in how to track brand mentions in AI, and the paid tools are compared in best AI visibility tools.
AI Overviews vs ChatGPT: Different Games
ChatGPT picks differently. When I asked it to recommend a lawyer, it built the list from Google Business Profile and Avvo listings (GEO for law firms). When I asked it for an AI consultant for an accounting firm, it named me first, from a plain, priced service page (how to rank in ChatGPT). Google's AI box, on the evidence above, leans on answer-first pages, the vendor's own docs, YouTube and Reddit. Same goal, different shelves.
If you want this run on your own buyer questions across Google's AI Overviews, ChatGPT and Perplexity, with the ranked list of what to fix, that's my AI Search Audit.
What does AI say about your business?
I'll run your domain through the citation checker I use on client work: where AI engines mention you, and the 3 fixes that matter most.
Get my free visibility reportFrequently Asked Questions
- How do I rank in Google AI Overviews?
- Google's own guidance (Search Central, 'AI Features and Your Website', read November 18, 2026) says there are no extra requirements: the page has to be indexed and eligible to show a snippet. In practice, the pages I saw cited answered the exact question in their first lines, often under a label like 'Quick Answer' or 'Short answer', with the number or the price right there. Google also runs extra related searches behind each AI Overview, so a page can be cited for a question it doesn't rank in the top 10 for.
- Do you need to rank on page one to be cited in an AI Overview?
- Not in my test. Across three buyer questions on November 18, 2026, the AI Overviews cited 29 sources and none of them were the pages holding the top 10 blue links underneath. Google's documentation explains why: AI Overviews may use 'query fan-out', running several related searches and pulling supporting pages from all of them. Ranking still helps you get found, but the AI box is picking from a wider pool.
- Why isn't my page cited in the AI Overview even though it ranks?
- It happens to me. My Claude Team plan page averages position 6.9 for 'claude team plan pricing' (Search Console, October 19 to November 16, 2026) and opens with the per-seat price, and the AI Overview still cited eight other sources, including two pages from Anthropic's own help center. Ranking and an answer-first opening make you eligible; they don't guarantee the pick. Check that the page can show a snippet (no nosnippet or tight max-snippet tags), then make the answer as specific and checkable as you can.
- Do YouTube and Reddit help with AI Overviews?
- For some questions, a lot. On 'how much does an AI receptionist cost', 6 of the 11 sources Google's AI Overview cited were YouTube videos (4) and Reddit threads (2). On 'how much does an AI consultant cost' it was 1 Reddit thread out of 10. If your buyers ask a cost or how-to question, a short video that states real numbers is a page Google can cite.
- Can I stop Google from using my content in AI Overviews?
- Google says the controls are the same as for regular snippets: the nosnippet and max-snippet robots tags and the data-nosnippet attribute. Blocking Googlebot in robots.txt removes you from Search entirely, not just from AI features. Most businesses want the opposite, so check these tags aren't set by accident on your money pages.
More on AI Visibility
Perplexity SEO (2026): How I Went From 2 to 12 Cited Answers in 6 Weeks
Perplexity SEO is getting Perplexity to cite your page and name you when buyers ask who to hire. I track the same 14 buyer questions every Monday. On October 5, Perplexity cited my site on 2 of 12. On November 16 it was 12 of 14, and it named me on 8. Here's exactly what changed, and why Claude still cites me on zero.
Ahrefs Brand Radar vs Semrush AI Toolkit (2026): Prices and One Real Bill
Ahrefs Brand Radar now starts at $50 a month for prompt tracking ($199 for its 578-million-prompt index), and Semrush's AI Visibility Toolkit is $99 a month for 25 prompts. Tracking the same 50 buyer questions on ChatGPT and Google's AI Overviews costs about $60 on Ahrefs, $159 on Semrush, $95 on Peec and $79 on LLMrefs.
GEO for Law Firms (2026): What ChatGPT Actually Uses to Pick a Lawyer
GEO for law firms means getting ChatGPT, Perplexity and Google's AI answers to name your firm when someone asks for a lawyer. I ran two real client questions through ChatGPT: every recommended firm came from its Google Business Profile and Avvo listings, and the website only supplied the reason to call. Here's what to fix first.
We Timed 283 New Pages on 3 Sites. Google Showed Most in a Day. One Site Waited Weeks.
I measured how long Google took to show 283 new blog posts on three sites I run. On two, 97% or more appeared within a week. On the third, half hadn't appeared after two weeks. Here is the data, the method and the check you can run.
Written by
Justin McKelvey
I spent 12 years running snowboard and mountain-bike shops in Colorado, then taught myself to code. Since then: 50+ products shipped, Head of Product at a national sports platform, CTO at an AI startup. Now I put AI to work in real businesses in Austin, TX and remote, and write down what actually works.
Work with meWant more help? Get your free AI visibility report · Book a strategy call
Before you go
Find out what AI says about your business.
Free personalized report: where ChatGPT and Perplexity mention you, where they recommend competitors instead, and the 3 fixes that matter most.
Get my free visibility report →The AI Visibility Starter: the DIY kit for getting recommended by ChatGPT.
Get started →