Find possible firms
“personal injury lawyer Sydney”
Lawyer · Founder · Researcher
How will people find lawyers in the age of AI?
I’m Dan Toombs, a lawyer and founder of FirmRanker. I’m studying how AI systems discover, evaluate and recommend law firms — and what that means for how people find and choose lawyers.
FirmRanker research
ChatGPTGeminiClaudePerplexity
01 The change
The shift
From search, to answer, to recommendation.
For most of the internet era, finding a lawyer online meant searching. A prospective client typed “personal injury lawyer Sydney” and received pages of choices.
AI changes the interaction. A person can describe the problem itself and ask which firms they should consider.
Traditional search presents choices. AI can increasingly narrow those choices — and sometimes make a recommendation.
That introduces a new layer between the prospective client and the law firm: the AI recommendation layer.
Narrow the field
“Who are the leading firms in Sydney?”
Make a choice
“I have this problem. Which firms should I consider?”
The further along this line a client goes, the fewer firms they see — and the more the AI’s view of your firm matters.
The commercial question
AI doesn’t have to replace Google to matter.
Legal services are unusual. A single instruction can represent substantial fees. AI only needs to influence a meaningful proportion of legal discovery for its recommendations to matter.
The old question
Does your firm rank?
The new question
Does AI know who your firm is — and does it consider you when someone asks for a lawyer like you?
For law firm leaders
What should a law firm do about this now?
The answer isn’t to panic. It also isn’t to wait until every uncertainty disappears.
The sensible starting point is to understand whether your firm is appearing in the AI systems your prospective clients may increasingly use — and what those systems appear to understand about you.
01
Measure
Establish a baseline across the markets and practices that matter.
With FirmRankerFree AI visibility check→02
Understand
See who appears instead and what sources surround the answer.
With Law By DanRead the analysis→03
Fix the foundations
Clarify expertise, people, locations, evidence and authority.
With Practice ProofGet help with the work→04
Experiment
Make changes, measure again and learn rather than assume.
With Practice ProofTest, measure, repeat→Start here
I’m a managing partnerUnderstand the commercial exposure and what to prioritise.AEO for Law Firms → I’m a practice managerFind out whether AI recommends your firm — in about a minute.Check your firm → I’m in marketingGet one practical, evidence-led idea every week.Join The AI Referral →Practical resource · Free to read
AI Visibility Checklist for law firms
Six areas to review before spending money on AEO — identity, expertise, corroboration, technical access and measurement.
02 The research problem
The new object of study
What happens between the question and the recommendation?
We cannot see every internal weighting. We can measure the information environment and the outputs.
Observable — recorded directlyInferred — studied from the outside
A The question
01
Client question
A person describes a legal problem in their own words.
Observable02
Intent
The system interprets what kind of help they need, and where.
InferredB The information environment
03
Retrieval
It may search the web or rely on what it already knows.
Partly observable04
Sources
What it reads — and what it chooses to cite.
Observable05
Firm entities
Names, lawyers, places and practices resolved into firms.
InferredSources may include
Firm websitesDirectoriesRankingsReviewsMediaRegulatorsProfessional bodies
C The answer
06
Consideration set
A short list of firms worth considering.
Observable07
Recommendation
Sometimes, an explicit choice.
ObservableWe should not pretend we know how every model weights every signal. We don’t.
That is exactly why the work starts with measurement before optimisation — recording what goes in and what comes out, repeatedly.
How we measure it→The work
I’m studying how AI chooses lawyers.
Recommendations
Which firms are actually surfaced and recommended?
Stability
Does the same question repeatedly produce the same firms?
Cross-platform differences
Do ChatGPT, Gemini, Claude and Perplexity agree?
Prompt sensitivity
Does wording or client intent change the consideration set?
Sources & authority
What does AI search, cite and use around law-firm answers?
Change over time
How stable is visibility as models and the web change?
Follow the research as it developsOne useful idea about AI and legal discovery, every week, in The AI Referral.
Join The AI Referral→03 The evidence
The research engine
Don’t guess. Measure.
FirmRanker treats AI visibility as something to be observed repeatedly, not as a permanent search ranking.
FirmRanker · How one observation is recordedMethodology illustration — no findings shown
1Ask
“I have this problem. Which firms should I consider?”
The same controlled prompt, by practice area and location.
2Across platforms
- ChatGPT
- Gemini
- Claude
- Perplexity
Each answer is preserved exactly as returned.
3Classify each firm
- Recommended
- Suggested
- Mentioned
- Not present
Plus its position and the sources cited or searched.
4Repeat
Run again, and again over time. Visibility is a distribution, not a screenshot.
Ask four AI systems the same question.
Same legal need. Same market. Different consideration sets.
FirmChatGPTGeminiClaudePerplexity
Firm ARecommended—SuggestedRecommended
Firm BSuggestedRecommended——
Firm C—SuggestedRecommendedSuggested
Illustrative example — not FirmRanker research data.
Controlled prompts
Comparable questions across practice areas and locations.
Repeated observations
Equivalent conditions repeated to measure variability.
Immutable raw answers
Original AI answers are preserved.
Entity resolution
Different names for the same firm resolved without changing the original answer.
Source analysis
Sources cited in the answer separated from sources merely searched.
Human validation
Automated extraction checked against reviewed ground truth.
04 The perspective
The perspective
I’ve spent more than 20 years inside this problem.
AI is new. The underlying problem isn’t.
I’ve worked across law, legal publishing, law reform, technology, legal marketing and law-firm growth. Each shift changed the tactics. The deeper question stayed remarkably consistent.
“How does someone who needs legal help decide which lawyer to trust?”
About Dan→Lawyer · Winston Churchill Fellow · Founder of FirmRanker & Practice Proof
I’ve spent more than two decades looking at legal discovery from different sides.
My perspective comes from more than one part of the legal industry. I’ve worked as a lawyer, legal publisher, law-firm growth adviser and founder of businesses involved in how people find, evaluate and choose legal services.
- Practice ProofLaw-firm growth and implementation.
- FirmRankerAI visibility research and measurement.
- Best SolicitorsLawyer discovery and evaluation.
- Mediations AustraliaConsumer-facing legal service discovery.
The writing
What I’m learning.
All analysis→05 Follow the work
The conversations
Law By Dan Podcast
Conversations about law, technology, legal business and the changing way people find and choose lawyers.
The AI Lawyer Discovery ProjectNew series in development
A newsletter by Dan ToombsWeekly · Free
The newsletter
The AI Referral
One useful idea about AI and legal discovery. Every week.
No generic AI news. No breathless predictions. Just research, observations and practical implications.
- The ideaOne sharp proposition
- What we observedThe evidence behind it
- Why it mattersWhat it means for law firms
- What I’m watchingOne open question
Free. Unsubscribe any time.