Newsletter Your Honor, We Need Clients newsletter cover: You Don't Have One AI Visibility Problem. You Have Five.

Digital Marketing & Content

You Don't Have One AI Visibility Problem. You Have Five.

By Sagar Pratap SinghPublished 21 Aug 2026Last updated 30 Sept 2026

Four AI engines share only 3.8% of the sources they cite, and directories are cited first in 77.8% of Perplexity legal answers. Why law firms must track AI visibility engine by engine.

Full issue

The AI engines now recommending lawyers disagree about which sources to trust, and the gap is wide enough that a single "AI visibility score" tells you almost nothing. Treat AI discovery as a set of separate contests, measure each engine on its own, and put your first effort into the third-party sources those engines read, because that is where they learn who you are.

A marketing director at a mid-size personal injury firm typed "best car accident lawyer in Chicago" into ChatGPT, then into Perplexity. Two lists, barely any names in common. She assumed she'd made a mistake. She hadn't.

The engines read different sources

Writesonic ran 161,286 prompts through ChatGPT, Gemini, Perplexity and Google AI Overviews in May and June 2026. Only 3.8% of cited sources were cited by all four, and 72 to 73% were unique to a single engine. ChatGPT and Perplexity shared about 13%. Forward Push, logging 38,173 answers across 25 practice areas and 15 cities, found ChatGPT's and Perplexity's lists of named firms for the same city and practice overlapped only about 6%.

Gemini shows how unstable this is: Citorian found it named a firm in 69% of personal injury answers, the highest of five engines, while Forward Push, weeks earlier, found about 12%, the lowest.

Directories are the gatekeepers, but not the same ones on every engine

The demand is real. BrightLocal found 45% of US adults now use AI for local business recommendations, up from 6% a year earlier, and Clio found 28% of consumers who asked AI a legal question were told to contact a lawyer.

The machine is making the referral, and when it does, it cites directories. InterCore ran 540 queries three times each on Perplexity across six practice areas and 30 metros: a directory was the first source in 77.8% of answers, and 34.2% of the firms named had no page of their own cited at all. In Citorian's data, Super Lawyers and Justia supplied 53% of citations. Below that top pair, the ranking moves: Lawyers.com took 5.72% of citations in InterCore's run and 0% in Citorian's, while Reddit took 8% in one and 2.04% in the other.

Advertising doesn't buy a place. Morgan & Morgan spent about $218.2 million on ads in 2024, per ATRA's analysis of Vivvix data, and barely featured in Citorian's answers. Citorian's own reading: "ad spend alone does not decide it." Paid search, meanwhile, averages $131.63 per legal lead in WordStream's benchmarks, double the cross-industry $66.69.

The leaders win everywhere; the middle is where engines split

Citorian's ten most-cited firms were named by all five engines. Named-partner trial firms with deep directory presence, such as Block O'Toole & Murphy (43% of New York answers) and Clifford Law Offices (43% in Chicago), win on every engine. The divergence punishes the firms in the middle, which is most firms. Phrasing matters too: vetting questions produced a named firm 94% of the time, emergency questions like "I was just hit by a car, what do I do" only 18%.

Distrust the numbers that arrive with the pitches

A June 2026 press release claims 78% of legal queries trigger a Google AI Overview, citing Semrush. Semrush's 10-million-keyword study says nothing of the kind: the overall rate was under 16% by November 2025, with Science highest at 25.96%. Ahrefs measured Law & Government at 23.1% against a 20.5% baseline. 5W's own April report had said 23.6%. The trend that holds up is quieter. Semrush found AI Overviews on commercial-intent queries up 71% between November 2025 and April 2026, and Ahrefs measures a 58% lower click-through rate for the top result when an Overview appears.

What to do this week

Measure per engine. Run ten queries a real client would type, both vetting and emergency, on each engine separately. Log mentions per engine, never averaged.
Claim the free surfaces first. Complete Justia and claim Avvo. Understand Super Lawyers: twelve indicators, peer evaluation, about 5% of a state's lawyers, and selection can't be bought.
Fix entity consistency. Name, address, phone, practice focus and bar details should match across Google Business Profile, directories and bar listings.
Check the right robot. GPTBot is OpenAI's training crawler; OAI-SearchBot governs ChatGPT search. Google-Extended doesn't touch Search; AI Overviews answer only to snippet controls such as nosnippet, which also cost you ordinary snippets.

Then ask the uncomfortable question: if a model had to verify you exist and are good, using only sources you don't control, what would it find?

Sources

Read the original on LinkedIn Share on X

Related service

Digital Marketing & AI Visibility

See how I help law firms