How to Vet Lead-Gen Counsel With Public Court Data

Every operator eventually asks some version of this: who should we actually retain? The usual answer is a name someone heard at a conference, or a directory badge on a firm's website. Neither tells you whether that firm has ever defended a case shaped like the one you will get.

There is a better answer, and it is sitting in public court records. It will not tell you who is "best" — nothing will — but it will tell you, objectively, who actually does this work, how much of it they do, and whether they do it on your side of the caption.

Why the directories cannot answer this

Chambers, Legal 500, Best Lawyers, Super Lawyers. They differ enormously in rigor — Chambers runs real client and peer interviews; others are closer to paid placement — but they share one limitation that matters here. None of them are outcome-based. They measure reputation among lawyers, which is a different thing from experience defending autodialer claims for a lead buyer.

They are useful for disqualifying and useless for ranking. Treat a directory listing as a floor, not a signal.

The dataset that can answer it

Federal civil litigation is public. PACER is the system of record, and it charges per page. The Free Law Project's RECAP Archive, surfaced through CourtListener, mirrors what people have already purchased from PACER and exposes it through a free API.

The important part for vetting: federal dockets carry a cause code — the statute the case is brought under. TCPA claims are filed under 47:227, the cause code for 47 U.S.C. § 227, Restrictions on Use of Telephone Equipment. FCRA claims — the statute behind trigger leads and prescreened offers — are 15:1681.

That means you can ask a precise question instead of a keyword question. Searching the phrase "Telephone Consumer Protection Act" returns anything that merely mentions it. Filtering on cause 47:227 returns cases actually brought under it.

CourtListener's search API also returns structured firm names on each docket, not just free text. So a firm-scoped query is a single request.

Every number you get is a floor

This is the part to internalize before you draw any conclusion. RECAP does not contain all of PACER. It contains what someone has already bought. Coverage is a by-product of what researchers, journalists and litigants happened to purchase.

So a firm showing 61 TCPA dockets has at least 61. It does not have exactly 61, and a firm showing 3 has not necessarily handled 3. Read every count as "at least N."

Six numbers worth pulling

Ordered by how much signal they carry per unit of confounding:

• Docket concentration. What share of the firm's federal filings are consumer-statute cases versus general commercial litigation? Specialization is measurable. "We have a TCPA practice" is not.
• Side. Do they appear for defendants, plaintiffs, or both? Plenty of firms work both sides of consumer statutes. That is not disqualifying, but you want to know before you share facts, and it is a conflict question worth asking out loud.
• Repeat clients. The same defendant appearing across multiple matters is the closest thing to a satisfaction metric in public data. Clients vote with renewals.
• Recency. Volume in the last twenty-four months separates an active practice from a page on a website. The law here moved substantially after the one-to-one consent rule was vacated; experience from 2018 is not the same asset.
• Venue overlap. Do they appear in the districts where you would actually be sued? TCPA filings concentrate in a handful of courts, and familiarity with the local bench is real value.
• Early termination. How often do their cases end on a motion to dismiss or a voluntary dismissal rather than grinding on? For a lead-gen defendant, cheap disposal is usually the product being purchased.

Getting side out of the free tier takes one extra step: the attorney-to-party linkage lives on CourtListener's parties endpoint, which requires a free API token. The search endpoint works without one.

The one number to ignore

Win rate.

Case difficulty is not randomly assigned. A firm that takes hard, high-exposure class actions with bad facts will post a worse record than one handling single-plaintiff demand letters with clean consent records. Raw outcome rates measure which cases a firm accepts at least as much as how well they handle them.

The inversion nobody mentions

Here is the structural limit of all of this, and it cuts against the whole approach.

Litigation data can only see counsel who ended up in litigation. The lawyer who fixed your opt-in language, tightened your vendor reps and warranties, and built your revocation workflow — so that you never got sued — leaves almost no docket trail at all.

The dataset systematically undercounts the thing a compliance-minded operator most wants to buy. Docket analysis is a strong filter for defense counsel and a weak one for preventive counsel, and those are frequently not the same people or even the same firms. Use the data to build the litigation shortlist. Use references and a work sample to judge the advisory side.

The free layer nobody checks

Court records show who cleans up. Regulatory filings show who is in the room while the rules are written.

The FCC's Electronic Comment Filing System is public and searchable. CG Docket No. 02-278 is the TCPA docket, and it holds every comment filed on consent, revocation and robocall rulemaking — with the filer and the client they filed on behalf of. The FTC's Telemarketing Sales Rule proceedings sit on regulations.gov the same way.

A protocol you can run in an afternoon

1. Build the shortlist from docket concentration and recency, not from directories or conference visibility.
2. Check side and conflicts. Do they appear for plaintiffs in consumer statutes? Do they represent lead buyers you compete with?
3. Check venue overlap against where your calling program actually exposes you.
4. Run the ECFS search for regulatory footprint.
5. Only then read the directories, and only to disqualify.

Everything above is public and costs nothing but time. If you want the analysis without assembling it yourself, the commercial legal-analytics platforms — Lex Machina, Docket Alarm, Bloomberg Law — package it with far better coverage than the free archive, and they are worth the subscription if you are making a real retention decision.

What to ask once you are in the room

The data gets you a shortlist. The meeting is where you learn whether it is the right one. Three questions worth more than the rest:

• "Walk me through a case you lost." The explanation is more diagnostic than any win rate. You are listening for whether they understood the exposure early and told the client plainly.
• "Who actually does the work?" The partner in the pitch is often not the person who will draft your consent language. Ask for the associate's docket history too — you can look that up the same way.
• "What would you change about our funnel before we ever get sued?" This one tests the preventive muscle that litigation data cannot see. A firm that only knows how to defend will answer it thinly.