Nobody has published the odds that ChatGPT cites the page ranking first on Google for the same legal query. The 1-in-15 figure that circulates in law firm marketing appeared on an agency page with no method, and the nearest real study measures something else: what share of ChatGPT's citations rank in Google's top 10.
Earlier posts on this site compared overlap across five engines and explained why a firm ranking first can lose the AI Overview. This one tests a single claim about ChatGPT and position one, shows why the published numbers cannot answer it, and gives a firm a way to measure it on its own page-one keywords.
- No published study measures the odds that ChatGPT cites the exact page ranking first on Google for the same legal query. The circulating 1-in-15 figure came from an agency page with no method attached.
- The closest published number, a Semrush figure reported by Custom Legal, says 6.82% of ChatGPT citations rank in Google's top 10. That is the share of citations that rank, not the share of top-ranking pages that get cited.
- Martindale-Avvo reported in April 2026 that ChatGPT's answers match Google's top 10 for legal queries less than 25% of the time, against roughly 75% for Perplexity and Claude.
- Ahrefs data reported by LEXGRO found ChatGPT reads Reddit in 67.8% of the pages it retrieves but cites Reddit in 1.93% of answers, so being read and being cited are separate events.
Where does the 1-in-15 claim come from?
The claim appeared on a Direct Online Marketing page about optimizing a law firm for AI, logged in Firm AEO's September 2026 research file with the note that no method was published. When Firm AEO re-checked the page on November 13, 2026, the figure was no longer in the page text, so the claim has no source a firm can verify today.
One in 15 is 6.7%. That sits very close to the 6.82% Semrush figure that Custom Legal reports, which suggests the claim may be that number restated with its meaning flipped. The agency did not say so, so treat the link as Firm AEO's inference, not a documented origin.
What do the published overlap figures actually count?
Each published figure counts a different unit, and none counts position one. Semrush, as reported by Custom Legal, counts ChatGPT citations that rank in Google's top 10. Martindale-Avvo counts answers that match Google's top 10. Ahrefs counts Google AI Overview citations, a different engine. A firm cannot convert any of them into odds for its own first-place page.
The table lists each figure with the question it answers. The last column is the one that matters for a firm asking about its own rankings.
| Figure | Source | Engine | What it counts | Answers the rank-one question? |
|---|---|---|---|---|
| About 1 in 15 | Direct Online Marketing (agency page) | ChatGPT | Not stated | Unknown, no method published |
| 6.82% | Semrush, via Custom Legal | ChatGPT | Citations that also rank in Google's top 10 | No, it starts from citations |
| Under 25% | Martindale-Avvo, April 2026 | ChatGPT | Legal answers matching Google's top 10 | No, top 10 and answer level |
| About 75% | Martindale-Avvo, April 2026 | Perplexity and Claude | Legal answers matching Google's top 10 | No, different engines |
| About 38% | Ahrefs, March 2026 | Google AI Overviews | AI Overview citations ranking in the top 10 | No, different engine |
Why is the share of citations that rank not the same as the odds of being cited?
The two questions run in opposite directions. The Semrush figure starts with pages ChatGPT cited and asks how many rank in the top 10. A firm's question starts with a page ranking first and asks how often ChatGPT cites it. The first can be small while the second is large, or the reverse, depending on how many sources each answer cites.
An illustrative example shows the gap. If ChatGPT cited eight pages per answer and one of them was always the first Google result, the share of citations that rank first would be 12.5%, while the odds of the first result being cited would be 100%. Real numbers are neither, which is why the claim needs its own test.
Does ChatGPT read the first-ranked page even when it does not cite it?
Often it may, and reading is not citing. Ahrefs data reported by LEXGRO found ChatGPT reads Reddit in 67.8% of the pages it pulls but names Reddit in only 1.93% of answers. A firm's first-place page can be in the material ChatGPT reads and still be left out of the sources it shows.
Martindale-Avvo's April 2026 analysis also found the sources ChatGPT cites most for legal queries are Super Lawyers, Avvo, Martindale-Hubbell and FindLaw. Directory profiles repeating a firm's facts can carry the firm's name into an answer without the firm's own URL appearing at all.
Is a first-place Google ranking still worth protecting?
Yes, though its value has shifted. Ahrefs found in February 2026 that position-one click-through rate on keywords with an AI Overview fell from 0.073 to 0.016, about 58% lower. The ranking still feeds the indexes that Perplexity, Claude and Gemini retrieve from, which overlap with Google far more than ChatGPT does.
The practical reading is that rank one is an input to several engines and a weaker input to ChatGPT. A firm should keep the ranking and also make sure the page carries a self-contained answer and that its directory profiles agree with it, since that is the material ChatGPT tends to show.
How can a firm measure the odds on its own pages?
Take the keywords where the firm ranks first in Google Search Console, rephrase each as the question a client would type, and run it in ChatGPT with search on, three times in a logged-out window. Log three separate outcomes, because the page, the domain and the firm name are cited at different rates.
Firm AEO treats a difference of one or two answers as noise, since ChatGPT output varies by day and by phrasing. Any public statement about the results, such as a claim to be the firm ChatGPT recommends, must be verifiable and not misleading under ABA Model Rule 7.1, and the firm's ethics counsel decides the wording.
- 1. Export queries where the firm's average position is 1.0 to 1.5 over the last 90 days.
- 2. Keep 10 to 20 that a client would plausibly ask a chatbot, rewritten as full questions.
- 3. Run each question three times in ChatGPT with search on, logged out, on different days.
- 4. Record whether the exact ranking URL is cited, whether any firm URL is cited, and whether the firm is named.
- 5. Divide each count by total runs. Those three rates are the firm's own answer to the 1-in-15 question.
Is the 1-in-15 figure wrong?
It cannot be judged right or wrong without a method. The closest published figure counts the share of ChatGPT citations that rank in Google's top 10, which is a different question from the odds that a first-place page is cited.
Does the same gap apply to Perplexity and Google AI Mode?
Less so for Perplexity, which Martindale-Avvo found matches Google's top 10 about 75% of the time on legal queries. Google AI Mode draws on Google's own index, so a firm should run the same three-outcome test there rather than assume either engine behaves like ChatGPT.
Can a firm advertise that ChatGPT cites it?
Only with care. AI answers change by day and by user, so a claim like that may be unverifiable or misleading under ABA Model Rule 7.1. Firm AEO is a marketing company, not a law firm, and the firm's ethics counsel decides whether and how to say it.
- 1.Direct Online Marketing, How to optimize my law firm for AI (agency page; 1-in-15 claim logged September 2026, not found on re-check November 13, 2026)
- 2.Custom Legal Marketing, ChatGPT optimization for lawyers (reports Semrush 6.82% figure)
- 3.Martindale-Avvo, AI visibility for law firms: an expanded guide (April 2026)
- 4.Ahrefs, How many AI Overview citations rank in the top 10 (March 2026)
- 5.Ahrefs, AI Overviews reduce clicks, updated study (February 4, 2026)
- 6.LEXGRO, Reddit, AI search and law firm marketing (reports Ahrefs Reddit retrieval data)
- 7.ABA Model Rule 7.1, Communications Concerning a Lawyer's Services