Walk into the aftermath of a bad storm and insurance becomes brutally physical. There are wet floorboards, missing shingles, a contractor with a tape measure, and a family asking when the house will be livable again. Somewhere behind that scene, numbers begin to move. The price of drywall in that ZIP code. The labor hours for replacing a roof. A prior claim attached to the property. A weather record confirming the hail. Many of those numbers travel through Verisk.
The Jersey City company is easy to misread because it does not fit one modern category. It is an enterprise software vendor, but it also maintains insurance forms and actuarial content. It is a climate-risk company, but its customers use the models to set premiums and buy reinsurance. It is an AI company, increasingly, but the valuable part is the specialist data and rules beneath the model. Verisk is best understood as decision infrastructure for insurance.
A utility learned to behave like a software company
Verisk traces its history to 1971, when Insurance Services Office began operations by consolidating several rating bureaus. The mission was practical: collect statistics, develop insurance programs, calculate prospective loss costs, and help property-and-casualty carriers navigate state regulation. There was no lone founder in a garage. The industry itself assembled the institution it needed.
That cooperative origin still matters. Insurance works better when carriers can use common definitions, comparable data, and policy language that regulators recognize. ISO became one of the keepers of that shared grammar. Over time, it added claims databases, property information, and analytics. The acquisition of AIR in 2002 brought catastrophe modeling; Xactware in 2006 brought repair estimating. Verisk Analytics became the holding company in 2008 and listed on Nasdaq the following year.
The IPO contains a revealing footnote. Existing shareholders - many of them insurers - sold roughly $1.9 billion of stock, making it the largest U.S. public offering of 2009. Verisk received none of the proceeds. A business created to serve insurers had become valuable enough for those same owners to cash out, yet its products remained threaded through their daily work.
Three expensive questions, packaged a hundred ways
Almost every Verisk product answers a variation of three questions: What can go wrong? What might it cost? What should we do next? ISO forms and rules help a carrier create the product. Property data and loss costs support underwriting and pricing. Catastrophe models translate hypothetical hurricanes, earthquakes, floods, fires, freezes, terrorism, and cyber events into financial loss distributions. ClaimSearch matches loss histories and flags patterns that may deserve investigation.
Then there is Xactimate, the name many homeowners never hear even when its numbers shape their settlement. Adjusters and restoration contractors use it to build room-by-room repair estimates from local material, equipment, and labor prices. XactAnalysis routes and monitors those assignments. ClaimXperience lets policyholders and professionals collaborate remotely. One damaged kitchen can therefore produce an entire Verisk procession: weather confirmation, claim matching, photographs, measurements, an estimate, a quality review, and reporting.
On the life-and-annuity side, FAST helps carriers launch products and administer policies on a cloud platform. The 2025 purchase of SuranceBay added producer licensing, onboarding, appointments, and compliance. The portfolio may look eclectic from outside. From inside an insurer, it resembles a map of the work.
- Weather becomes a loss curve
- A building becomes replacement cost
- A claim becomes a pattern
- A regulation becomes a workflow
- A portfolio becomes capital demand
Recurring revenue with an unusually deep memory
Verisk generated $3.07 billion of revenue in 2025. Underwriting contributed about $2.18 billion; claims produced about $893 million. More important than the mix is the contract structure: annual subscriptions and long-term agreements represented more than 80 percent of revenue. Customers generally prepay quarterly or annually for continuous access to hosted products and updated content.
2025 revenue mix
The moat is not simply a large database. Data ages, and competitors can buy imagery, build models, or hire actuaries. Verisk's advantage comes from several pieces reinforcing one another: decades of contributed insurance records, specialist teams, regulatory relationships, accepted industry content, and products embedded in customer systems. Usage generates feedback and, in some cases, more data. Better data improves the product. The improved product becomes harder to remove.
That does not make the company untouchable. Moody's RMS and Karen Clark & Company compete in catastrophe modeling. LexisNexis Risk Solutions overlaps in claims and fraud data. CoreLogic's successor businesses and a growing field of aerial-imagery and property-data specialists chase underwriting and claims work. Guidewire, Duck Creek, and Applied Systems own other layers of insurance software. Large carriers can build internally. Verisk's answer is breadth with domain depth: connect more of the decision chain without pretending one application can replace the carrier's core system.
Science under pressureHow to sell an honest estimate of the unknowable
Catastrophe modeling is the most vivid expression of Verisk's expertise. A model invents thousands of plausible event years, simulates hazards at individual locations, estimates how buildings respond, and applies insurance terms to calculate losses. Verisk's updated U.S. winter-storm model uses a 10,000-sample-year climate catalog. Its 2026 tropical-cyclone model adds finer exposure detail and richer building characteristics. Synergy Studio, launched in 2026, puts the model suite on a cloud-native platform with one financial framework.
The output can look precise - an average annual loss, an exceedance curve, a number of dollars - while the future remains uncertain. Good modeling does not erase uncertainty; it organizes it. That distinction is becoming more important as inflation changes reconstruction costs, development moves into exposed regions, and climate conditions strain historical assumptions. Insurers need numbers they can put into pricing and capital decisions, but they also need to know where the numbers bend.
Verisk sits downstream from climate science and upstream from a commercial decision. Its models matter when they change a deductible, a reinsurance purchase, a concentration limit, or the price of coverage.
The laboratory is also a workplace of unusual neighbors. Across roughly 8,000 employees, actuaries sit beside meteorologists, physicists, insurance lawyers, GIS specialists, construction-price researchers, and software engineers. Verisk says its culture is organized around results, learning, and caring; its 2025 employee engagement score reached 80 percent. The company also earned U.S. Great Place to Work certification for a tenth consecutive year. For a knowledge business, retaining that mix is not an HR footnote. It is part of the production system.
The next layerAI arrives with a filing cabinet attached
In 2026, Verisk brought selected analytics and generative AI capabilities into Anthropic's Claude. It also began promoting Coraline, a set of human-in-the-loop agents for insurance product design, regulatory monitoring, and maintenance. The strategic idea is straightforward: a general model becomes more useful in insurance when it can draw on specialist content, understand the workflow, show its reasoning, and leave accountable experts in control.
Verisk has a plausible advantage here because its value has always been context. A carrier does not merely need text generation. It needs to know that a court decision may require policy language to change, which state filing is affected, how an exclusion relates to a line of business, and which reviewer must approve the result. AI can compress the clerical work. The proprietary data, legal interpretation, and audit trail remain the product.
The company also has a live reminder that strategy does not unfold neatly. Verisk agreed in 2025 to buy roofing-contractor software provider AccuLynx for $2.35 billion, then tried to terminate the deal after an extended regulatory review. In August 2026, Delaware's Chancery Court ordered it to resume commercially reasonable efforts to seek approval. Verisk said it disagreed and was weighing its options. The episode puts a hard edge on its ambition to connect the claims network from insurer to contractor.
For customers, the practical pitch is less dramatic. Use Verisk to launch an insurance program faster, sharpen a property estimate, test a portfolio against an event that has not happened, detect a suspicious relationship among claims, or reduce the manual work between a loss and a repaired home. Verisk's place in the market is not at the glossy front door of insurance. It is in the engine room, where a scientific estimate, a legal clause, and a construction price must agree long enough for someone to make a decision.