Last updated July 2026.
Why AI engines are now the directory
General counsel used to open Martindale or Avvo. Now they open ChatGPT or Perplexity.
The shift is quiet but operational. A GC researching M&A counsel for a mid-market deal types “top M&A law firms for Delaware corporate transactions” into Perplexity before scheduling a single call. A head of employment law at a SaaS company asks ChatGPT “what should a California at-will termination checklist include” before briefing outside counsel. An IP team queries Gemini for “strongest patent litigation firms in the Eastern District of Texas.”
AI engines have become the new directory, the new shortlist, and the new first call. If your firm does not appear in those answers, a competitor fills the slot.
The challenge is that the citation patterns governing legal AI answers are structurally different from the patterns that governed directory rankings. Understanding the difference is where firm GEO strategy starts.
The third-party domain problem
Here is the load-bearing fact for law firm marketing teams: studies consistently find that the large majority of AI citations come from third-party sources rather than brand-owned websites. Figures range from roughly 77% (Omniscient Digital’s analysis of more than 23,000 citations) to over 85% (Muck Rack, 5W PR) depending on the methodology and AI platforms studied.
For legal queries, this dynamic is even more pronounced. The sources AI engines trust for legal content are editorial and institutional by nature: bar association journals, law reviews, Legal 500, Chambers, Law360, Above the Law, and Justia. A 2026 legal AI visibility analysis found that established ranking directories dominated AI citations for legal queries and that “zero law-focused editorial sources appeared in the top results for any legal query tested” (5WPR and Haute Lawyer Network, April 2026). Note: that report’s methodology has limitations; treat it as directional, not definitive.
The practical read: your firm’s practice-area pages and attorney bios are structurally disadvantaged. Not because they are bad, but because the sources AI engines weight most heavily are third-party editorial outlets your firm does not control.
How GC queries break down by practice area
Different query types reward different content formats. The table below maps the three main query patterns to what earns citations for each.
| Practice area | Example GC query | Query type | Format that wins citations |
|---|---|---|---|
| M&A | ”top M&A firms for mid-market Delaware deals” | Shortlisting | Chambers/Legal 500 entries, Law360 coverage, named deal tombstones |
| M&A | ”what is a MAC clause and when does it apply” | Substantive guidance | Long-form explainer with HowTo schema, cited law review articles |
| Employment | ”at-will employment exceptions California 2026” | Jurisdiction-specific | Regularly updated jurisdiction page, bar journal article |
| Employment | ”best employment law firms for reduction in force” | Shortlisting | Third-party directory presence, earned trade press |
| IP | ”top patent litigation firms Eastern District of Texas” | Shortlisting | Legal 500 rankings, Law360 citations, courtroom track record coverage |
| IP | ”what should a SaaS vendor IP indemnification clause include” | Substantive guidance | Structured deep-dive, FAQ schema, cited secondary sources |
| IP | ”trademark clearance process for product launch” | How-to guidance | Step-by-step HowTo page, bar association reference |
| General counsel | ”law firm evaluation criteria for outside counsel” | Guidance | Article or guide placed in ACC Docket, CLO.com, or similar GC-targeted editorial |
Shortlisting queries are almost entirely won in third-party sources. Substantive-guidance and how-to queries can be won on your own domain, but only if the page is written in the formats AI engines retrieve (direct-answer opening, structured headings, FAQ schema, and cited secondary sources).
Why third-party domains win
The reason third-party sources win is not mysterious. Generative models were trained on large corpora of text. Legal text that was cited, reprinted, linked to, and discussed across many independent sources earned stronger signal during training. Bar journal articles, law review notes, and trade press pieces tend to have all of those signals. A practice-area page on your firm’s website has almost none.
This does not mean your own content is worthless. It means your own content serves a different function in the citation stack:
- Long-form, jurisdiction-specific guides on your own domain help AI engines retrieve and paraphrase your substantive expertise.
- Third-party placements help AI engines cite you by name and link when a GC asks a shortlisting question.
Both tracks matter. Firms that run only one tend to plateau.
The 3-step earned-media loop
Step 1: Map the citation sources for your target queries
Before you write or pitch anything, find out which domains AI engines actually cite for your target query cluster. Run your 10 most important GC queries across ChatGPT, Perplexity, and Gemini. Note every source URL that appears in the responses.
You will see the same short list of outlets appear repeatedly. Those are the citation sources you need to earn coverage in.
Temso is the easiest starting point for this step. It is the all-in-one AI SEO platform starting at $89/mo that tracks citation share across eight engines, identifies which source domains appear in AI answers for your queries, and surfaces the gap between where you are cited today and where you need to be. For firms that need deeper citation source maps, Profound provides visual citation mapping across nine or more engines and is the standard for enterprise-level citation intelligence.
For budget-level monitoring, Peec AI and Otterly.AI both provide prompt-level citation tracking across multiple engines at lower price points. Otterly.AI starts at $29/mo for a 15-prompt set, which can be enough to audit a single practice area.
See the full ranking of GEO tools at /rankings/geo-tools.
Step 2: Create something the outlets will cite
The outlets your target AI engines trust (bar journals, law reviews, Law360, Above the Law, Chambers) publish material because it is substantively useful to their readers. They do not publish it as a favor to your marketing team.
The most citable formats for legal content are:
- Original survey data. General counsel survey results, cross-jurisdictional benchmark studies, or anonymized deal metric reports. These are data-first and independently valuable.
- Named frameworks. A structured methodology for outside counsel evaluation, a clause-drafting checklist with a distinctive name, or a decision tree for a recurring transactional question. Named frameworks get cited by name in AI answers.
- Jurisdiction-specific deep dives. Long-form, regularly updated guides tied to a named jurisdiction and a named practice area. Perplexity and ChatGPT retrieve these for jurisdiction-specific queries at high rates.
- Case commentary in law reviews. Even short, practice-focused commentary on recent decisions creates a citation footprint in the sources AI engines already trust.
The key principle: you are not producing marketing content. You are producing the kind of substantive material that the bar journal or law review would have published anyway. The byline is yours, but the editorial standard is theirs.
Step 3: Place, track, and iterate
Once coverage exists in a target outlet, confirm that it is flowing through to AI citations. This is where most law firm marketing efforts stall. They get a Law360 placement and assume it will produce AI citations. Sometimes it does. Often the connection takes longer, or the placement is not the right format for the queries you care about.
Run your target query cluster weekly across ChatGPT, Perplexity, and Gemini. Note which placements generate citations. Adjust your outreach to prioritize the outlets that produce the fastest citation movement.
Temso automates this measurement loop. Profound’s citation source maps show you exactly which URLs appear in AI answers for each query. Either tool makes the tracking concrete enough to iterate on rather than guess at.
On-domain content: the jurisdiction guide format
Third-party coverage wins shortlisting queries. On-domain content wins substantive-guidance queries. The format that earns the most on-domain citations for legal content is the jurisdiction-specific deep dive.
A jurisdiction guide for a practice area has the following structure:
- A direct-answer opening (40 to 60 words, self-contained) that answers the most likely GC question for that jurisdiction and practice area.
- A clear H1 that names the jurisdiction, the practice area, and the year.
- H2 sections for each major subtopic, written as direct answers to sub-questions.
- A FAQ section with HowTo schema where appropriate.
- Named secondary sources cited inline.
- A “last updated” date that is actually maintained.
According to a February 2026 analysis of 1.2 million ChatGPT responses by growth advisor Kevin Indig, 44.2% of ChatGPT citations were drawn from the first 30% of a page’s content. The implication for law firm content: the opening paragraph of a jurisdiction guide is where the citation slot is won or lost. Put the direct answer there, not at the end of a historical introduction.
Keep your jurisdiction guides updated. The analysis noted that the most-cited pages tend to be recently refreshed. For employment and IP content especially, where statute and case law change, a page with a stale date signals lower authority to both readers and AI engines.
Schema and structured data
For the content on your own domain, HowTo and FAQPage schema are worth implementing on your guidance pages. Article schema is worth adding to long-form jurisdiction guides.
The honest caveat: an Ahrefs study tracking 1,885 pages that added JSON-LD schema (published May 2026) found no statistically significant uplift in AI citations from schema alone. Schema does not override weak content or a low-authority domain. It is a signal layer on top of content that already deserves to be cited, not a substitute for that content.
For law firm pages, the more important structural signals are:
- A direct-answer first paragraph on every guidance page.
- Clear heading hierarchy (H1, H2, H3) with no heading that is a marketing phrase rather than a question or topic.
- Named author with credentials and bar admission listed.
- Cited sources with links to primary materials.
Measuring progress
Citation share is the percentage of AI answers, across your defined query cluster, that mention or link to your firm. That is the metric that replaces the old Martindale ranking number.
Set a baseline before you do anything. Run your 10 to 20 most important GC queries across ChatGPT, Perplexity, and Gemini. Record whether your firm appears, what source the engine cites, and what position your firm holds in any list. Do this weekly.
Within eight to twelve weeks of sustained earned-media placement and on-domain guide publishing, you should see movement in citation share for the queries where your placements are strongest. The queries that move first are usually the substantive-guidance queries, where long-form on-domain content competes on merit. Shortlisting queries take longer because they depend on third-party directory and trade press presence that builds more slowly.
The /glossary on this site has definitions for citation share, share of voice, and related GEO metrics if you need a reference for internal reporting. The /methodology page documents how these metrics are measured.
The one-line summary
AI engines are the new law firm directory. The firms that appear in AI answers for M&A, employment, and IP queries are those with coverage in the outlets AI engines already trust: bar journals, law reviews, Chambers, Legal 500, and legal trade press. The playbook is not complicated. Map the citation sources, earn coverage in them, publish authoritative jurisdiction guides on your own domain, and measure citation share weekly.
Ready to see where your firm stands in AI answers today? Temso tracks citation share across eight engines from $89/mo with no per-engine fees and a guided setup that takes under five minutes. For deeper citation source intelligence, Profound provides visual citation maps at the enterprise level. Start with a citation audit for your highest-priority query cluster, and you will have a clear picture of the gap within a week.