Your Google Rankings Mean Nothing to AI Search

Last updated: 2026-06-01

A year ago, 76% of AI Overview citations came from websites ranking in Google’s top 10. Today that number sits between 17% and 38%, depending on the study. Moz found that 88% of Google AI Mode citations go to pages outside the organic top 10 entirely. If you have spent years getting your practice site onto page one and assumed that was enough to stay visible when patients or clients search with AI, this week’s data says otherwise.

This is not a minor adjustment to how search works. It is the biggest structural shift since Google became the default way people find local professionals. And most Canadian solo practices have no idea it happened.

What Google announced at I/O 2026

Google replaced the traditional search box with an AI interface at its I/O conference last week. The new system is built on Gemini 3.5 Flash and runs through what Google is calling “information agents,” which monitor the web continuously rather than returning a static list of links. AI Overviews now reach 2.5 billion monthly users. AI Mode, which gives fully conversational responses to search queries, crossed one billion monthly users at the same event.

In plain terms: more than a billion people are now getting answers from AI rather than clicking through to websites. Google AI Overviews appear in 25.8% of all US searches today, and that share is rising each quarter. The AI is not showing them a ranked list. It is composing an answer and citing the sources it trusts, which may or may not include your site, regardless of where you rank.

Why rankings and citations now run on different tracks

Google’s organic ranking algorithm and its AI citation system use different signals. Ranking is primarily about backlinks, page authority, and keyword relevance accumulated over years. AI citation is about whether the AI can extract a clear, trustworthy answer from your page at the moment the question is asked.

A site can rank well for “family dentist Hamilton Ontario” and still be invisible in AI answers because the page does not directly answer the questions AI users are asking. Conversely, a site that ranks on page three can get cited regularly if it has clean structured data, a published FAQ, and consistent information across third-party directories. The 73% of page-one brands with zero AI mentions are not doing something wrong in the traditional SEO sense. They are just optimised for a system that no longer determines who gets found.

What actually drives AI citations

Three things account for most of the difference between practices that get cited and practices that do not.

Structured data is the most controllable factor. Pages with FAQ schema in place see 2.8 times higher citation rates across major AI platforms, according to a 23-study meta-analysis published earlier this year. Schema-rich pages overall are 2.5 times more likely to appear in AI-generated answers. This is not because AI reads schema the way Google once read meta keywords. It is because structured data gives the AI a machine-readable version of your claims, which it can verify and extract with confidence.

Earned media and third-party validation account for 85.5% of all AI citations in professional categories. Your own website is one input. What Google, Yelp, your provincial regulator, and industry directories say about your practice carries far more weight in the AI’s source selection. A chiropractor cited in a local news article or a lawyer listed with detailed specialties on their provincial bar website will outperform a competitor whose only presence is a polished practice site.

Entity consistency matters more than content volume. AI cross-checks the information it finds across multiple sources. If your name, address, specialty, and credentials say one thing on your website and something slightly different on your Google Business Profile, the AI either hedges or defaults to the third-party source it trusts more. Consistency across every directory, regulator listing, and review platform is the foundation everything else builds on.

Each AI engine cites on its own logic

Not all AI platforms cite the same sources, which means practices need to think about where their authority lives, not just whether their website is accurate.

ChatGPT cites Wikipedia for 47.9% of its top-10 references. If your name or your practice type is described correctly in Wikipedia’s coverage of your profession, you benefit from that association. ChatGPT also favours clean heading structure and direct-answer formatting on the pages it does retrieve. It cites only about 15% of the pages it actually visits.

Perplexity leans heavily on Reddit, which accounts for 46.7% of its top citations. For regulated professionals who cannot post to Reddit directly, third-party discussions of your specialty and consumer review threads still feed Perplexity’s picture of who is trustworthy in your area.

Google AI Overviews pull 23.3% of citations from YouTube. For a practice that has published even a handful of short, well-described videos, that is a meaningful advantage over competitors who have not. The transcript and description text, not the video itself, is what AI reads.

Only 11% of domains are cited by both ChatGPT and Perplexity. Appearing in one does not mean appearing in both. A full AI visibility strategy has to account for where each engine looks.

Four fixes a Canadian solo practice can make this month

The gap between a practice with good AI visibility and one with none usually comes down to a handful of decisions that take hours, not months.

First, add LocalBusiness and Organization schema to the website. This is a small block of JSON code placed in the page source. It tells AI crawlers your name, address, phone, hours, and service area in a format they cannot misread. A developer or a capable web designer can add it in under an hour.

Second, publish a FAQ page with five to seven direct-answer questions about your practice. These should cover what you do, where you are, who you serve, what a first appointment involves, and what sets your approach apart. Answer the questions in the first sentence of each response, then add context. This format is the most extractable structure for every major AI platform.

Third, audit your information across directories. Check that your name, address, and phone number match exactly on your website, your Google Business Profile, your provincial regulator’s directory, and any other directory where you are listed. Inconsistencies are the single most common cause of AI getting a practice wrong or skipping it.

Fourth, request reviews that describe what you do specifically. A Google review that says “great service” does nothing for AI visibility. A review that says “helped me understand my options for a corporate restructure in two consultations” is the kind of earned-media signal AI citation systems weight heavily. Manta AEO’s review coaching guide covers exactly how to ask for these.

Frequently Asked Questions

Does this mean SEO is now worthless?

No. Traditional SEO still builds the domain authority that AI citation systems partially rely on, and organic search still drives traffic for queries where AI does not generate an answer. The problem is assuming that a good SEO ranking alone will transfer to AI visibility. For most small practices, that assumption is no longer correct.

How do I know if AI is actually citing my practice?

You can test it manually by asking ChatGPT, Claude, Perplexity, and Google AI Mode about your profession in your city, then checking whether your name comes up. For a systematic picture, a citation audit tracks mentions across platforms and shows exactly which queries you appear in, which you are missing from, and how that compares to nearby competitors.

Does this apply to smaller Canadian cities, not just Toronto and Vancouver?

Yes, and in some ways it matters more in smaller markets. AI platforms have thinner data on mid-sized cities. A practice in Kelowna, London, or Fredericton that builds its AI visibility correctly faces less competition for citation share than a practice in a major market. The gap between visible and invisible is often just a few hours of structured work.

My website is new. Do I need to wait for Google to rank it before AI can find me?

No. AI citation and search ranking develop on different timelines. A new site with clean structured data, a well-written FAQ, and consistent directory listings can start appearing in Perplexity and ChatGPT answers within weeks, before it ranks on page one for anything. This is one of the few areas where a newer practice can compete immediately with one that has been around for a decade.

Do I need to be on Reddit and YouTube to have AI visibility?

Not personally. For regulated professionals, direct participation on those platforms is often impractical or restricted. The citation patterns from Reddit and YouTube reflect where general internet authority lives, not where your marketing budget should go. Structured data on your site and consistent earned media from directories and local press will move your AI visibility without requiring a content creator role.

---

Seventy-three percent of practices ranking on Google’s page one have zero mentions in AI-generated answers. If your website was optimised for the last decade of search, it almost certainly needs work to be visible in the next one. A free AI visibility audit at manta-aeo.com shows exactly where you stand across ChatGPT, Perplexity, Claude, and Google, with specific fixes ranked by impact.