POWERADMIN AI
IndustryJune 5, 2026·11 min read

Mass Tort Firms Are Built to Sign Cases. They Are Not Built to Keep Them.

A mass tort partner walked me through his pipeline: 14,200 active plaintiffs across four bellwether dockets, $8 million a year in lead acquisition, a 22-person bilingual intake team, a 41% sign-up conversion rate. Impressive machine. Then I asked what percentage of signed plaintiffs he loses before resolution. The answer: about 19%.

Do that math. Roughly 2,700 cases, at an average net fee of $11,000 per plaintiff, is about $29 million in fees quietly walking out the back door while the front door gets all the budget. Mass tort firms are built to sign cases. They are not built to keep them.

The silent phase nobody automates

A mass tort case has three phases. Phase one, acquisition: ad spend, lead vendors, call centers, retainers, fact sheets. Firms put 80% of their operational budget here and run it with real playbooks. Phase three, resolution: disbursement, liens, checks. Established processes. Phase two is the problem: the 14 to 36 months between sign-up and resolution, filled with discovery, bellwether trials, science day rulings, and expert depositions, none of which is visible to the plaintiff, because nobody tells them anything.

Three forces drive the 19% attrition:

Why standard fixes do not work

Monthly newsletters. Generic MDL updates open at 18 to 24% in month one, 8% by month four, effectively zero by month six. Plaintiffs don't care about the docket. They care about their case.

Quarterly call-center outreach. The vendor's script has no access to the case file, so it can't answer anything real. Plaintiffs hang up more frustrated than before, and the phone numbers keep dying between calls.

"We'll call when there's news." The most common model and the worst one, because during the silent phase there is almost never individual news. So there is almost never contact. So plaintiffs leave.

Hiring a plaintiff liaison. At 14,000 plaintiffs, one liaison doing nothing else would need 670 touches a month. Three liaisons cost $250,000 a year and still can't cover it.

Notice the pattern: every fix treats retention as a broadcast problem or a staffing problem. It's neither. It's a per-plaintiff, per-case-event, per-channel coordination problem, which is precisely the shape of problem software should own.

What the AI Fusion build looks like

Three coordinated agents inside one AI Super Agent, sharing state through a knowledge layer connected to your CMS (Litify, Filevine, Smart Advocate), phone system, SMS platform, email, e-signature, and fact sheet portals.

The Contact Health Agent (always on)

Monitors three signals per plaintiff: last successful engagement (open, reply, click, or pickup), last plaintiff-initiated contact, and carrier validation of the phone number via API lookup. At 75 days without engagement, it sends a soft, case-specific re-engagement text (these open at 64 to 71%). At 120 days with no response, it escalates: number verification, USPS address cross-reference, and if both fail, a paralegal flag with a report of what was tried.

The Case-Event Update Agent (trigger-based)

Fires on CMS changes, PACER bellwether feeds, and internal milestones: sample selection, deposition scheduling, expert filings, MSA posting. It drafts updates personalized with the plaintiff's name, their claim facts from intake notes, and their case number, not a docket bulletin, and delivers by the plaintiff's preferred channel. A Communications Supervisor reviews every message for compliance, accuracy, and privilege risk before it goes anywhere, and routes ambiguous items to a paralegal. The metric that moves: plaintiffs receiving a substantive, case-specific update within 90 days goes from under 25% to 92 to 97%.

The Inbound Response Agent

When a plaintiff calls asking "is my case still alive?", the agent answers in seconds with the case file, docket status, last firm action, and pending requests in front of it, and updates changed phone numbers on the spot. Generic chatbots resolve 12 to 20% of these calls. This build resolves 78 to 84%, and the rest escalate to paralegals with the full context attached: withdrawal talk, expert deposition questions, anything that needs a human.

The math on retention

Back to the original firm: 14,200 plaintiffs, 19% attrition, $11,000 average net fee. Deployments like this cut silent-phase attrition by 50 to 65%; call it conservatively 19% down to 9.5%. That's roughly 1,349 cases retained, or about $14.8 million in recovered fees, against a deployment cost in the low six figures annually plus one or two retention paralegals. North of 70x over the life of the docket.

There's a second-order effect too: settlement administrators calculating MSA disbursement pay attention to which firms have documented, engaged, contactable plaintiffs. Firms full of non-responsive claimants take haircuts. Firms with clean contact health get paid first and better.

What this is not

Not a chatbot: a chatbot can't pull docket status from your CMS or write a case-specific update. Not legal advice: the agent surfaces facts and routes strategy questions to humans, and the supervisor enforces that line. Not a CMS replacement: it sits on top of Litify or Smart Advocate with read access to case data and write access to communications. And not one-size-fits-all: every MDL gets its own configuration (questionnaire schedules, science day timing, bellwether calendars) during the deployment month.

The implementation playbook

Retention metrics start moving by day 90 and compound by day 180.

The takeaway

Lead acquisition is commoditized and everyone's costs are rising together. The next advantage isn't signing more. It's keeping what you signed, while competitors lose 19% of their plaintiffs to silence. Ask your case managers for your attrition number today. If nobody knows it, that's the answer. If it's over 12%, you have a workflow problem that costs eight figures over the life of a docket.

Want the retention math run on your specific docket?

Book a 20-minute call. We will pull the contact-health audit and project the recovered case value before you sign anything. Or start with Voice AI free.

By Harry Hedaya, Founder, Power Admin AI

Want to see this on your own operation?

Book a 20-minute working session and bring a real workflow or your real numbers. We'll show you exactly what an AI build would do with them, and if it's not a fit, we'll say so.