A class-certification order issued last month by the U.S. District Court in Boston is a practical warning for any insurance agency that buys internet leads, retains outsourced marketing vendors, or runs automated calling and texting campaigns. The decision does not resolve whether Liberty Mutual violated the Telephone Consumer Protection Act. It does something that, for the defense, can be worse: it lets the case proceed on behalf of tens of thousands of alleged call and text recipients, and it does so by treating the validity of consent as a question that can be answered for everyone at once.
In Adam Ward v. Liberty Mutual Insurance Company, Judge Brian E. Murphy certified two TCPA classes on June 12, 2026. For independent agencies, the order’s significance lies not in Liberty Mutual’s ultimate liability, which remains to be decided, but in how the court viewed the lead-generation chain — and in how little that familiar multi-vendor structure did to keep the dispute out of classwide territory.
The Lead Chain at Issue
The order describes a chain that will be recognizable to anyone who buys leads. Liberty Mutual ran telemarketing campaigns using leads obtained from third-party aggregators, which in turn sourced them from various websites. It engaged a separate platform, Jornaya, to document consent by capturing playback evidence for each contacted consumer.
The leads at issue were purchased from All Web Leads, Inc., which had sourced them from Next Level Media, LLC. Between March 17 and June 12, 2020, Next Level Media sold All Web Leads 24,587 leads drawn from a single website, www.instant-auto-insurance-now.com. Those leads were compiled into a spreadsheet and sold on to Liberty Mutual, which then used yet another vendor, Drips Holdings, LLC, to place prerecorded calls and send texts.
One entity captures the inquiry, another aggregates and resells the lead, a third documents the consent record, and a fourth makes the contact. That division of labor is the industry’s standard operating model. Ward shows that it is not, by itself, a shield: where the calls and texts were made by or on behalf of the insurer whose products were marketed, the layered chain did nothing to prevent classwide treatment.
What Ward Alleged
Adam Ward registered his number on the National Do Not Call Registry in 2005. Between March 26 and March 30, 2020, he received six calls from Liberty Mutual — three using a prerecorded voice — and one text message. His number had been submitted through the lead website to Liberty Mutual on March 26, 2020, but Ward denied submitting the form and denied giving express written consent, or any consent, to be solicited. Whether he is right is a merits question the court did not reach. The only question before it was whether the case could proceed as a class action.
The Two Certified Classes
Judge Murphy certified a prerecorded-voice class and a do-not-call class.
The prerecorded-voice class covers persons in the United States or its territories to whom Liberty Mutual placed, or caused to be placed, one or more calls between March 1 and June 30, 2020, to a cellular number, where Drips’s records indicate a prerecorded message played and the number appears on the All Web Leads spreadsheet. The do-not-call class covers persons who received more than one call in a twelve-month period by or on behalf of Liberty Mutual, on a number that had been on the National Do Not Call Registry for at least 31 days, during the same window, where the number appears on the same spreadsheet.
On numerosity, the court noted more than 20,000 potential prerecorded-voice members and more than 7,000 potential do-not-call members.
Why Consent Became a Classwide Question
For agencies and producers, the heart of the order is its treatment of consent. Liberty Mutual argued that consent demands an individualized, lead-by-lead review. The court rejected that argument at the certification stage, holding that consent here turns on a common form: whether the lead website could constitute consent to be contacted by Liberty Mutual is a question common to the entire class. As the court framed it, either the law required Liberty Mutual to be named on the consent form — defeating its defense across the board — or it did not, allowing the defense to be proven across the board. Either way, the consent defense rises or falls in a single ruling.
That is the point insurance marketers should sit with. In a purchased-lead case, the instinct is to argue that every lead is different. Ward holds that where the alleged consent flows from the same website or a substantially uniform form, the legal sufficiency of that mechanism is a common question fit for classwide resolution. A vendor’s spreadsheet, consent token, or compliance report may be evidence. It is not the end of the inquiry.
The Damages Arithmetic
Certification matters because TCPA exposure is measured per call and per text. The statute authorizes up to $500 for each violation and up to $1,500 for each willful or knowing violation — $500 and $1,500 under Section 227(b)(3) for the prerecorded-voice claims, and the same figures under Section 227(c)(5) for the do-not-call claims.
The order’s class figures are members, not violations, and the distinction cuts against the defense. Each prerecorded-voice member received at least one offending call; each do-not-call member, by definition, received at least two. So 20,000-plus prerecorded-voice members and 7,000-plus do-not-call members describe a floor on the violation count, not a ceiling — and because the do-not-call claim rests on multiple calls per person, the per-violation tally, and the exposure, run higher than a one-per-member count would suggest. At $500 per violation, a class of this size implies eight figures of baseline exposure before a willfulness finding trebles it. These are not damages awards; they are the arithmetic that explains why certification, not the merits, is often the decisive event in TCPA litigation.
The Massachusetts Overlay: Chapter 93A
Ward is a federal TCPA case, but for a Massachusetts agency the federal statute is only half the picture. The TCPA contains no fee-shifting provision; class counsel ordinarily look to the recovery fund rather than a separate fee award. Massachusetts law removes that limitation.
The Commonwealth maintains its own do-not-call regime under M.G.L. c. 159C and the telephonic-solicitation rules at 201 CMR 12.00, and those rules are promulgated under Chapter 93A. A telemarketing practice that violates the state regime is therefore an unfair or deceptive act under Chapter 93A — which carries the double and treble damages and the mandatory attorneys’ fees that the federal statute lacks. Conduct that violates the TCPA will frequently violate the state regime as well, and the same automated insurance campaign that produces a federal prerecorded-voice claim can open the Chapter 93A door alongside it. For a Massachusetts agency, in other words, the exposure is not capped at the TCPA’s per-violation figures, and the “no fee-shifting” comfort that applies in federal court does not travel to Chapter 93A.
Buying Leads Is Not a Safe Harbor: Braver
The lead buyer, not just the lead generator, can end up holding the liability. Braver v. NorthStar Alarm Services, LLC is the cautionary case. There, the lead seller — Yodel Technologies — made the calls using soundboard technology; NorthStar bought the resulting leads. The seller’s lack of consent was not disputed. The question was whether NorthStar, which never placed a call, could be held vicariously liable for Yodel’s violations. The court held that it could, under all three federal agency theories — actual authority, apparent authority, and ratification — resting on a set of facts that describe an ordinary lead-buyer arrangement: NorthStar reviewed and effectively approved the script, knew soundboard technology would be used, set the lead-delivery method, directed calls to ZIP codes where it did business, adjusted the pace and procedures, and shared campaign data with the seller.
The figures are the warning. Roughly 252,765 calls over nine months yielded NorthStar a mere 150 customers — and exposure the court’s reasoning placed in the range of $126 million to $379 million. A small, largely unsuccessful campaign produced potentially existential liability, on the theory that coordinating how leads are generated can make the seller’s conduct the buyer’s own. The lesson is blunt: buying leads rather than generating them does not remove TCPA risk, and the more an agency shapes the script, targeting, cadence, or transfer process, the more it looks like a principal.
The Soundboard Through-Line — and Why AI Voice Changes Nothing
The Braver facts are not a historical curiosity. They are the first link in a chain that runs straight to today’s AI-voice question.
After Braver, NorthStar and Yodel asked the FCC to declare that soundboard technology — a live agent selecting prerecorded audio clips in real time — was not an “artificial or prerecorded voice” under the TCPA, precisely because a human was choosing each snippet. In its 2020 Soundboard Ruling the FCC refused, holding that a live agent’s presence does not negate the statutory prohibition; a call that plays prerecorded audio is a prerecorded-voice call regardless of the human at the controls. That ruling cites Braver by name.
In February 2024, the FCC built directly on the Soundboard Ruling to confirm that AI technologies generating human-sounding voices — voice cloning and the like — are “artificial or prerecorded voice” under the TCPA and require prior express consent absent an emergency or exemption. The reasoning is the same one NorthStar and Yodel lost: it does not matter that a call sounds individualized, interactive, or human. If the voice is artificial or prerecorded, the statute applies.
That continuity is the takeaway for any agency now experimenting with AI voice tools, automated callbacks, or conversational outreach platforms. The argument that an AI agent is lifelike enough to fall outside the prerecorded-voice rules is a newer dress on the soundboard argument that has already failed twice at the Commission. And the regulators did not stop at declarations: the FTC pursued Yodel itself, extracting a civil penalty and a telemarketing ban, and the company’s operations did not survive. The vendor that built the test case is gone; the liability theory it spawned is not.
What Agencies Should Take From the Order
Ward does not mean every purchased lead creates liability. It means lead generation and automated outreach are compliance-controlled functions, not merely marketing ones — and that they deserve the same discipline a Massachusetts agency already brings to licensing, premium handling, carrier appointments, and claims reporting. Before relying on a vendor list, an agency should be able to answer:
- Who generated the lead, and through what website or form?
- Did the consent language identify the agency, carrier, or producer that would actually make the contact?
- Was the number scrubbed against the federal and Massachusetts do-not-call registries?
- Was the outreach a live call, a text, a prerecorded message, or an AI-generated voice contact?
- Who holds the consent record, and can it be produced in usable form if challenged?
- Does the vendor contract address TCPA and Chapter 93A compliance, indemnity, audit rights, data sources, and proof of consent?
- Is the agency controlling the script, geography, cadence, transfer process, or follow-up in a way that could make the vendor’s conduct its own?
For Massachusetts agencies the risk is not theoretical. The Ward order came out of the federal court in Boston, in a personal-lines marketing campaign, and it sits atop a Chapter 93A regime that supplies multiple damages and fee-shifting the federal statute does not. The campaign in the case was larger than the average agency’s, but the operational question is identical: if an agency cannot prove valid consent at the source, a purchased lead is not just a sales opportunity. It is evidence.
