Know what every law firm is doing, every week.
We read the public pages of 1,200+ law firms every week and turn them into structured data you can query. Practices, people, matters, clients, hiring.
Sarah Chen advises rare earth processors on cross-border transactions. She recently represented a Canadian producer in its $1.2bn acquisition of a lithium developer, and counsels several clients on Section 232 exposure from the Washington DC office.
The dataset
A structured record of the market, refreshed weekly.
We crawl the firms that bill the overwhelming majority of large-firm revenue worldwide, extract every signal type they publish, and resolve it into one graph.
Over a million structured records today, across more than twenty signal types per firm.
What it is worth
What it costs to decide without it.
Firms make these calls with the best information available today, which is a conversation, a CV and last year’s rankings. These are a few of the line items that sit behind them.
lost every year by the largest 400 US firms on lateral hires that do not work out.
the cost of a single failed lateral, depending on firm tier.
for a strategy engagement that is accurate the day it lands and stale by the next quarter.
a year to staff a small intelligence function that can cover roughly twenty firms by hand.
Sources: ALM Intelligence and Decipher Investigative Intelligence. One better-informed decision covers this many times over.
Who uses it
One dataset. Everyone who has to make a call about the market.
Eight teams to start with, and the list keeps growing as firms tell us what they ask it. The question changes depending on who is asking. The underlying record does not.
Management committee
- Which practice to stand up next
- Which office to open, and where
- Which rivals are converging on us
Practice group leaders
- Who else is building in my sector
- How deep is my bench against theirs
- Where is the market heading first
Business development
- Which clients appear on rival matter lists
- Where a client’s work is expanding
- Who is courting our relationships
Pitch and RFP teams
- Prove a capability is genuinely scarce
- Prepare against a named competitor
- Show depth behind a single partner
Lateral recruiting
- Who is in motion this quarter
- What a candidate actually worked on
- Which teams are quietly thinning
Marketing and content
- Which firms publish on our sector
- Who runs a podcast, and how often
- Where the share of voice sits
Competitive intelligence
- Track a comparator set continuously
- Get alerted when something moves
- Keep a live view, not a quarterly rebuild
Client relationship partners
- Which firms serve my client elsewhere
- What those firms are good at
- Which panels are being rebuilt
…and the teams we have not built for yet: pricing, knowledge management, alumni relations, opening a new office. If it depends on what the market is doing, it is the same data.
You cannot hire a practice into existence the month a client calls.
Partners take quarters to recruit and years to season. When every firm runs the same analysis through the same models, what separates them is whether the bench was already there. The firms that will lead the next sector started building for it some time ago.
What the data can do
At least sixteen things this data makes possible.
How it works
Extract, index, then compare against last week.
Extract
Every signal type on a firm’s site, turned from prose into structured records. Bios and matter descriptions are sentences, not fields, so they have to be read rather than parsed.
Index
Resolved into one graph. Lawyer, client, practice, matter and office, linked to each other and to a point in time.
Diff
Each pass compared against the last. What appeared, what moved, what quietly went away. Entities are compared, not pages, so a site redesign does not read as a thousand changes.
Why this has not existed
Two thousand bespoke sites
No shared schema between any two of them, and one redesign makes a firm look brand new.
The same lawyer, four ways
J. Smith, John Smith, Jonathan A. Smith Jr. Matching every record against every other is quadratic, so it has to be blocked to finish at all.
Why weekly matters
A single crawl tells you where the market stands. Repeating it turns a number into a direction.
Illustrative shape. We began recording recently, so the earliest part of any curve cannot be reconstructed, by us or by anyone.
From the current crawl
Things nobody had counted.
A first census of the market. No model can answer these, because answering them means visiting every site and counting.
practice pages have fewer than three attorneys attached. A page is not a practice, and now you can tell which is which.
client names appear on more than one firm’s site. The shared counsel map, drawn for the first time.
names for the same practice area. Energy Transition, Renewables, Power, Clean Tech, all one thing.
open roles name a specific practice area. Hiring intent, visible before anyone announces it.
firms publish a podcast. Nobody had counted, including the firms competing with them.
practice areas listed, median firm to top decile. Breadth claims, finally measurable.
Sourcing and security
Public sources only. Nothing of yours goes in.
Every record comes from pages any member of the public can open. There is no integration, no upload and no access to your systems, which is why there is usually nothing for IT to assess.
Company
Briefly has been reading legal experience data for years.
Telemetry is not a new direction. It is the same pipeline, pointed outward.
Submissions
We parse directory submissions for Chambers, Legal 500, WTR, IFLR and Benchmark.
From lawyers
Our tools let attorneys contribute their own matters, straight from the people who did the work.
An experience graph
Cross-linked across firms, directories and years.
The open web
The same extraction problem, applied to every firm rather than one at a time.
Submissions run ahead of websites.
A matter appears in a filing months before it reaches a practice page, because a deadline forces it out. That is where we learned to read this market, and it is why we could point the same pipeline at the open web.
Questions
Common questions.
How current is the data?
The crawl runs on a weekly cycle. Signals that carry their own publication date, such as client alerts, lateral announcements and job postings, are dated from the source rather than from the crawl.
How much history is there?
Not much yet, and we would rather say so. We began recording recently, which means the comparison layer is young. Nobody else has a longer record of this market either, because nobody was keeping one.
Can I limit it to the firms I actually compete with?
Yes, and this is the request we hear most often. You define a comparator set, the practices and markets that matter to you, and results are scoped to that rather than to the full crawl.
Do you need access to our systems?
No. There is no integration and no data flows from you to us. Everything is read from published pages, which is what usually keeps this out of a lengthy security review.
How do we actually use it?
We are shaping this with the firms we are talking to. The options on the table are a chat interface, a dashboard with alerts on a watched set, written briefings on a question you set, and an MCP endpoint so your own agents can query the dataset directly. Tell us which of those you would use.
Why can an AI model not just answer this?
A foundation model is a snapshot taken at training time. It knows what a firm published before its cutoff, it has not looked since, and it never stored two versions of anything, so it cannot tell you what changed. Counting how many firms have a given practice today means visiting those sites today.
Get started
Name a practice area. We will show you the market.
Tell us the sector and the firms you care about, and we will come back with what the current crawl says about them.
Public sources only. No firm data, no client data, no documents ingested.