Does ranking on Google get your law firm cited by AI?

The short answer: far less than you would hope, and it depends heavily on which AI you mean. We paired thousands of real Google results pages with the AI answers to the same questions on the same day. When a site ranked in Google's organic top 10, Google's own AI Overview cited it about a third of the time, Gemini about a quarter, and ChatGPT about one time in eight. Flip the question and it gets starker: most of what the AI engines cite, 69 to 83 percent depending on the engine, came from outside that day's top 10 entirely. Page 1 of Google and the AI answer to the same question are overlapping but genuinely different contests.

The setup

We track the questions New Yorkers type when they need a lawyer, across ChatGPT, Gemini, and Google AI Overviews. Because our AI Overview tracking captures the full Google results page, every one of those samples also records the classic organic links for the same search, from the same page load. That gives us something rare: not a rank tracker and an AI tracker glued together, but the ranking and the AI answer from the same question at the same moment.

Over seven weeks ending August 21, 2026, that produced 2,676 captured results pages, 73 percent of which showed an AI Overview. The rest showed no AI Overview at all, which is its own story. We normalized every organic link and every AI citation to its registrable domain, then asked two questions. First: when a domain ranks in the top 10, does each engine cite it in its answer to the same question that day? Second: of the domains each engine cites, how many were in that day's top 10?

Read that chart from a law firm's point of view. Earning a page-1 ranking is the hard-won prize of legal SEO, and on the day you hold it, the most likely AI outcome on every engine is that your site is not cited at all. Even Google, assembling an AI Overview directly on top of the results page you rank on, cites a top-10 domain in about one case in three.

Position 1 is the whole premium

Within page 1, position matters in a lopsided way. The site at position 1 gets cited at two to three times the rate of positions 6 through 10. But the drop is front-loaded: falling from position 1 to the 2-to-5 band roughly halves your citation odds on every engine, while the difference between position 5 and position 10 is comparatively small.

For AI Overviews the gradient is clean: 62 percent at position 1, sliding steadily down the page. That makes sense, because the Overview is built from Google's own retrieval. ChatGPT is the outlier: even the number-one ranked site for a question gets cited in its answer only 21 percent of the time. If your firm's AI strategy is "rank number one and the rest follows", it holds up half the time on Gemini and one time in five on ChatGPT.

So where do the AI answers come from?

If most citations are not the top-10 rankers, what are they? We flipped the join: for every domain an engine cited, we checked whether it was in that day's top 10 for the question it answered.

Some of that gap is benign: an engine can cite a deeper page from a site that ranks for a related search, or pull from position 11. But the size of the gap, even on Google's own surface, means the AI answer is assembled from a much wider pool than the ranking you fight over. Two details from the same data underline how different the pools are. Gemini, which leans toward big established brands, cited a ranked social or forum platform (Reddit, YouTube, Instagram) exactly zero times in 384 chances. And legal directories do not fill the gap either: ranked directories were cited at a lower rate than other ranked sites on every engine's AI answers except ChatGPT, where both rates are low.

It also depends on what kind of law you practice

The overlap is not one number. Pooling the two chat engines by practice area, the share of ranked domains that got cited spans an order of magnitude:

What this means for your firm

• Your SEO is not wasted, but it is not the same game.Position 1 still carries a real citation premium, especially on Google's own AI surfaces. What no longer holds is the assumption that page-1 visibility and AI visibility are the same thing measured twice. They correlate, weakly on some engines, and 69 to 83 percent of AI citations go to sites that were not in the top 10 that day.
• Check each engine separately. A firm can be well-cited on Gemini and invisible on ChatGPT from the identical ranking. The engines disagree about lawyers on the same question, and they clearly weigh Google's ranking differently when they pick sources.
• Being citable matters as much as ranking. The engines are drawing from a wide pool, and the sites they repeatedly cite look consistently different: substantial content, machine-readable markup, established domains. That is the part of the contest you can work on without fighting for one ranking slot.
• Measure the outcome, not the proxy. Your rank tracker is measuring the old contest. The only way to know whether AI engines cite you is to ask them the questions your clients ask, repeatedly, and watch. That is what we measure, every day, per engine.

Usual caveats: one metro, one seven-week window, same-day pairing by design (we deliberately do not count a citation this week against a ranking from last month), and registrable domains rather than exact pages. The AI Overview comparison comes from the same page load as the ranking; ChatGPT and Gemini answers are paired to the same question on the same day.