Ten years ago, somebody who needed a lawyer typed "lawyer near me" into Google, scanned the three pins on the map, pulled up two or three websites, and started dialing. The firm's job was to be visible on that map and look credible on the landing page. Calls followed.
That path is thinning. More prospects now describe their problem to ChatGPT, Gemini, or Google's AI Overview and ask which lawyers to consider.
The model answers in a short paragraph and names two or three firms. The map still exists. Fewer people get to it.
Pew Research found that when an AI Overview appears on a Google results page, users click a traditional search result just 8% of the time, versus 15% without one, and only 1% ever click a source link inside the Overview itself. A firm that never makes the AI's shortlist doesn't lose the pitch. The phone simply never rings.
The Prospect Who Asked the Chatbot First
This person opens ChatGPT, types two sentences about a DUI arrest, and asks what to do. The model walks them through the next 72 hours, flags the DMV hearing deadline, and when pressed for names, offers two or three firms with a line of reasoning about each. The prospect never opens a browser tab. They screenshot the answer, text it to a family member, and call the first name on the list.
The firm that got named here wasn't necessarily the one ranking first on Google. It was the one the model had enough citable material about to feel confident recommending. That's a different competition, and a lot of firms haven't entered it. That is the shift from the map pack to AI-curated lawyer shortlists, and the firms that win here are the ones whose expertise is legible to a language model, not just to a crawler.
The Prospect Who Checked the AI Against Google
A second type of prospect still runs the Google search, but uses it as a second opinion. They've already got two names from the AI. They look to see whether those firms show up in the map pack, scan a few reviews, and compare the shortlist against the paid ads and organic results.
The firms that convert this prospect are the ones that appear in both places. Being on the AI shortlist gets you considered. Showing up again on the search page gets you believed.
A firm that dominates traditional SEO but is invisible to the models gets flagged as suspicious: why didn't the chatbot mention them? A firm the AI loves with no footprint on Google reads as thin. Both signals have to line up.
The Prospect Who Never Searched At All
The third category is the quietest, and probably the largest. This prospect never typed a query. They described the situation in a thread, got an answer, acted on it, and moved on.
No search log. No referral header. No form fill. From the firm's analytics, this person simply does not exist.
Attribution dashboards built around clicks and sessions miss this traffic entirely. The call that lands at the intake desk looks like a word-of-mouth referral, because the prospect says a name and nothing else. Marketing teams staring at declining organic sessions can mistake this for a demand problem when it's a measurement one.
The Model Has No Reason to Name This Firm
Language models recommend what they can defend. A study of AI search visibility across major engines found that household-name brands appear in roughly 73% of relevant AI answers, while niche brands appear in about 11% — a steep visibility ladder that mirrors how often a model has encountered a name in authoritative contexts. For a law firm, "authoritative context" means a specific set of signals:
- Substantive written content. Pages that answer the questions prospects actually ask, in enough depth that a model can quote them. Thin service pages built for keyword density don't qualify.
- Third-party citations. Mentions in bar journals, local news, legal directories, podcast transcripts, and nonprofit explainers. The model reads these as corroboration.
- Named attorneys with public footprints. Bios, speaking appearances, published articles, case commentary. A firm whose lawyers only exist on the firm's own site is hard for a model to validate.
- Clean structured data. Practice areas, jurisdictions, and contact information marked up so machines can parse them without guessing.
Firms that never make the shortlist usually have a decent website and nothing to offer on any of the four. The model has nothing to work with, so it names someone else.
Change These Things Before the Next Quarter
The practical move isn't to abandon traditional SEO. The map pack still delivers calls, and will for a while. Add a second track that treats AI engines as a distinct audience with distinct reading habits.
Start by auditing what ChatGPT, Perplexity, and Gemini currently say when asked to recommend a firm for your practice area and jurisdiction. If your name doesn't appear, work backward: what would a model need to see to name you?
Write the pages that answer the questions prospects actually bring. Get your attorneys quoted somewhere other than your own blog. Fix the structured data. Re-run the prompts in 60 days and see whether the shortlist moved.
Firms that read this as a measurement problem rather than a visibility problem will spend the next year wondering where the calls went. The ones that do the work show up on the shortlist, and the phone starts ringing again.
