A 7.4% organic rebound, an AI story buried inside renewal pricing, and a roofing acquisition stuck in court — all at 7.5x revenue
VRSK · Earnings Call · 2026-11-05
Verisk Priced the Moat, Then Hid the AI
What actually changed this quarter
On the surface, Verisk's third quarter was the relief quarter management had pre-announced: 7.4% organic constant-currency growth with underwriting at 7.7% and claims at 6.7%, subscription revenue (84% of the mix) up 8.2%, adjusted EBITDA margin up 90bps to 56.7%, and adjusted EPS up 14% to $1.96. Guidance was maintained, not raised — revenue $3.19–3.24B, adjusted EBITDA $1.79–1.83B, EPS $7.45–7.75 — with the CFO steering expectations "towards the midpoint of the range." A rebound confirmed, not a rebound upgraded.
The more revealing change is in what Verisk chose to talk about. The quarter's keyword profile is not an AI story at all — it is a data-ownership story. The top of the list is sophisticated clients, followed by contributory data and emerging risks. Lee Shavel's answer to JPMorgan's Andrew Steinerman is the thesis statement, and it is a quietly contrarian one for anyone who assumed generative AI commoditizes data vendors:
“We've always said that our most sophisticated clients use a broader set of our data than the less sophisticated clients. AI is effectively increasing the data sophistication of all of our clients, but that has to go through a very careful calibration as the industry is adjusting to these data sets.” — Lee Shavel, President and Chief Executive Officer · 2026-11-05
That reframes AI from threat to demand multiplier — and the proof points are hard, not rhetorical. The excess & surplus dataset went from $18 billion of premium to $36 billion in a single quarter, on 21 data contributors. Digital Media Forensics went from 600 million images in March to over 800 million. The best anecdote is the top-five self-insured retailer that took workers' comp claims in-house expecting AI to carry it, found outcomes were worse than with Verisk's industry data, and returned to sign a large multiyear anti-fraud contract.
If it works on a single client's data set, then it works for all of them. That's the value of the breadth of the industry data and the quality and the organization, the calibration of those data sets on a consistent basis.
The market moved on; the company monetized it
Here is the contrast worth flagging. AI risk was a top-three keyword across the entire market in Q1 2026. By this quarter it has dropped out of the global top-75 entirely — attention moved to agents and physical AI. Verisk spent that interval building an actual product line around it: an AI litigation tracker, AI exposure mapping, a U.S. data center exposure database, and an emerging-issues workshop bundled inside the Core Lines subscription. The company is monetizing a theme the keyword tape has already stopped bidding. That is usually the more durable side of the trade.
The product cadence supports it. Synergy Studio launched in late June and shipped its first quarterly update on time and on budget, with 16 clients live. Mozart Form Composer added nine new customers this year, including the largest agreement Verisk has ever signed, with a Tier 1 carrier doing an enterprise-wide forms review. And the newest entry, Forms AI Builder, went from beta in mid-September to 14 companies using it.
But the AI revenue is invisible — by design
When Goldman's George Tong asked whether agent-based usage through the Claude MCP Connector is a separate revenue stream, the answer was clarifying:
“For the time being, that is included in our existing and primarily our core lines contract where the data is coming from. It is and will continue to be a driver of the strong price realization in those contracts.” — Elizabeth Mann, Chief Financial Officer · 2026-11-05
So the AI upside is buried inside strong price realization and renewal terms, not a reported line. That is why the quarter reads as "maintained guidance" rather than an AI inflection, and why the burden of proof shifted to renewal pricing and contract duration. Shavel conceded the quantification problem in July and it hasn't been solved: “I don't think that we are able to quantify certainly its current impact.” — Lee Shavel, President and Chief Executive Officer · 2026-07-29 Cost-side evidence is better documented — FAST Studio delivering up to 30% productivity gains across the software development lifecycle — but management was explicit that savings get reinvested, with 25–75bps of margin expansion as the Investor Day target rather than a windfall.
The weather tax, and the AccuLynx limbo
Two unglamorous items carry more of the risk than the AI narrative. First, the contractor and adjustor customer base — roughly 10% of consolidated revenue — is now visibly stressed by two consecutive light catastrophe years. A second quiet year of weather is a revenue headwind arriving at the exact moment Verisk wants to sell roofing and restoration connectivity. “The contractors, particularly small businesses are under pressure with that lighter volume, and we have seen 1 or 2 bankruptcies in the space.” — Elizabeth Mann, Chief Financial Officer · 2026-11-05 The adjustor clients keyword surfacing in the top-30 is new, and not a happy kind of new.
Second, AccuLynx. The Delaware Supreme Court order on October 23 affirmed the Chancery Court judgment, and the deal now sits with the FTC with no completion guarantee. Management gave the shortest possible answer on capital allocation: “There is no meaningful change in our capital priorities.” — Lee Shavel, President and Chief Executive Officer · 2026-11-05 Compare that with July 2025, when AccuLynx was the single highest-ranked keyword on the call and Shavel was selling a roofing TAM: “It's approximately a $2 billion TAM. And as we've conveyed, AccuLynx is looking at approximately $150 million in 2025 of revenue.” — Lee M. Shavel, President and Chief Executive Officer · 2025-07-30 A growth asset has become a litigation footnote while $700 million of buyback authorization and a 2.5x leverage position sit in the middle of the 2–3x target range.
The tape got there first
Verisk is up 15.1% over the last 90 trading days, but that follows a 34.2% drawdown from the June 2025 peak of $321.33 — and the de-rating was brutal. Price to revenue has fallen to 7.5x from a 14.9x peak and is down 49% year over year, and price to free cash flow sits at 19.8x, down 56% YoY. The multiple, not the business, absorbed the damage.
Revenue reached $806M in the last reported quarter, up 4% YoY within a 16-year trend of +184%. But the cash conversion wobbled this quarter: operating cash fell 34% and free cash flow 41%, on higher semiannual interest, higher cash taxes and collection timing, with net interest expense at $53M against $42M. Effective net cash stands at −$3.9B, down 51% YoY, and interest coverage has slipped to 6.9x from roughly 10x a year ago.
The tell to watch
Verisk is the rare case where the AI bull case and the AI bear case are the same sentence: the data is the product, and everything else is packaging. What resolves it is not another product launch but three observable things — whether longer sales cycles, now flagged for a third straight quarter, finally compress; whether data overages and transactional strength prove structural rather than a one-quarter services bump the CFO explicitly declined to repeat; and whether AccuLynx closes or dies. Maintained guidance toward the midpoint, at 7.5x revenue, is a reasonable price for a moat. It is not yet a re-rating case.