Verisk Analytics, Inc. (NASDAQ: VRSK) — A Wide-Moat Toll Booth Whose Toll Finally Got Cheaper

Date: June 6, 2026 Price (2026-06-05): $181.73 · 52-wk range: $155.94–$322.92 (≈44% below high) · Market cap: ~$23.8B · EV: ~$27.9B Fiscal year: December 31 · CIK: 0001442145 · Segment: single reportable segment (“Insurance”)

⚡ Claude’s Take

This block is the author’s own subjective opinion. It is general information, not investment advice. The body of this memo (sections 1–15 below) takes no position, sets no price target, and discusses valuation only as embedded expectations and scenarios.

Verdict: ACCUMULATE ON WEAKNESS — a wide-moat compounder whose multiple has finally stopped being absurd. Conviction: Medium. Constructive entry zone ~$160–185; base-case fair value ~$200–230 over 2–3 years; not a short at any price I can defend.

Tag: “The toll booth is fine — the toll on the toll booth is what got cheaper.”

Verisk is one of the genuinely great businesses in the public market: a regulated-data near-monopoly on U.S. P&C insurance that takes no underwriting risk, collects ~80%+ of revenue as prepaid auto-renewing subscriptions, earns ~70% gross / ~56% adjusted-EBITDA margins and ~36% ROIC, and raises price mid-to-high-single-digits every year into a client base that already includes all of the top-100 U.S. P&C insurers. For most of the last decade the only thing wrong with it was the price — it routinely changed hands at 30–40x earnings. That objection has now substantially eased: the stock has de-rated ~44% to ~21x forward earnings / ~18.5x EV-EBITDA, the cheapest quintile of its own ten-year history (19th–25th percentile), even though the franchise is executing inside its long-term framework (FY25 +6.6% organic revenue, +8.5% organic adjusted-EBITDA). The market is now pricing the low end of management’s 6–8% growth algorithm — so the bar to do well from here is modest: the franchise simply has to keep being itself.

What the market is genuinely worried about — and what keeps my conviction at Medium rather than High — is real, not imagined: organic growth has stepped down quarter by quarter (FY24 +7.2% → FY25 +6.9% → Q1-2026 +5.5%), the core U.S. underwriting market is saturated so growth is almost entirely price not units, and generative AI plus cheaper public data is a credible medium-term solvent for the commoditizable edges of the data stack (point property/auto data, basic estimating, document extraction). The bull case is that the deceleration is tough-comps-and-soft-weather cyclicality and the 6–8% algorithm re-asserts; the bear case is that ~5% is the new ceiling and pricing power eventually cracks. Valuation sits right on that fault line: today’s ~$182 is the low edge of my base case, with ~10–25% downside if the bear is right and ~10–20%+ upside (plus a ~5% FCF yield and mid-single-digit buyback) if the base holds. The single thing that would flip me bullish (high conviction): two or three quarters of organic growth re-accelerating back through ~7%, proving the deceleration was cyclical. The single thing that would flip me bearish: evidence that annual price escalation in the core ISO/underwriting franchise is meeting genuine customer resistance — that is the load-bearing wall of the whole thesis, and if it cracks, ~21x is not cheap. I would buy weakness toward the low-$160s/high-$150s with both hands; at $182 I would accumulate patiently and let the deceleration question resolve.

1. Executive Summary

Verisk Analytics is a pure-play data, analytics, and technology vendor to the global property-and-casualty (P&C) insurance industry. After divesting Financial Services (2022) and Energy/Wood Mackenzie (2023), it now reports a single “Insurance” segment generating $3,072.7M of FY2025 revenue (+6.6% organic constant-currency), of which ~80%+ is recurring, prepaid, auto-renewing subscription revenue. The business is split into two revenue categories — Underwriting (71%) and Claims (29%) — and sells decision-support tools embedded in the regulatory and operational workflows of insurers, reinsurers, brokers, and government agencies.

The investment question is not whether this is a good business — it plainly is — but whether a high-quality, slowing, single-vertical data monopoly is correctly priced after a 44% drawdown. Verisk’s moat is among the widest in the analytics universe: Greenwald’s strongest advantage class (economies of scale coupled with customer captivity), reinforced by regulatory incumbency (a licensed statistical agent in all 50 states; the author of industry-standard policy forms and loss costs across 32 lines) and a 50-year contributory-data monopoly (38.9B underwriting records, 1.9B claims records) that no entrant can finance into existence. The moat is visible in the financials: ~70% gross margin, ~44% GAAP operating margin (normalized), ~56% adjusted-EBITDA margin, ~36% ROIC, >1x FCF conversion, and durable mid-to-high-single-digit annual pricing power into a fully-penetrated client base.

The bear case is equally concrete: (1) the core U.S. underwriting market is saturated — all top-100 carriers are already clients — so growth is price, not units; (2) reported organic growth is decelerating (FY24 +7.2% → FY25 +6.9% → Q1-2026 +5.5%); (3) generative AI and cheaper public data threaten the commoditizable layers of the stack (the 10-K itself names “free or relatively inexpensive information” as a demand risk); and (4) the FTC’s blocking of the $2.35B AccuLynx acquisition (terminated December 2025) signals that the moat can no longer be widened cheaply by M&A and that antitrust authorities view Verisk’s market power warily.

On valuation, the de-rating is largely sector-wide (S&P Global, Moody’s, and FactSet all compressed on the same AI-disruption fear) rather than a VRSK-specific blow-up. At ~21x forward earnings Verisk sits between its closest structural analogs (SPGI ~19x, MCO ~24x) and at the 19th–25th percentile of its own decade. A reverse-DCF implies the market is underwriting only the low end of management’s 6–8% framework; scenario analysis frames a Bear zone ~$135–160, Base ~$185–230, and Bull ~$260–320, with the current price at the low edge of Base. This memo takes no recommendation and sets no price target; it concludes that Verisk is a structurally excellent, capital-light, ~36%-ROIC franchise whose principal open question is whether its organic deceleration is cyclical or structural — the answer to which determines whether ~21x is a bargain or merely fair.

2. Business Overview

What Verisk does. Verisk Analytics is a data, analytics, and technology vendor to the global P&C insurance ecosystem. Following the deliberate dismantling of its diversified-information-services strategy — the divestiture of Financial Services (2022) and Energy/Wood Mackenzie (closed February 1, 2023) — Verisk is today a pure-play insurance-analytics company. It collects and processes billions of records, builds proprietary statistical models, and sells the resulting decision-support tools — embedded in client workflows — to insurers, reinsurers, brokers, and government agencies that need to price, underwrite, and settle insurance risk (FY2025 10-K, “Our Company,” p.4). In 2025 its clients included all of the top 100 U.S. P&C insurance providers for the lines it serves — a fact that is both a moat tell and a saturation warning.

Segment structure. Verisk reports a single operating segment, “Insurance” (FY2025 10-K, Note 19), a simplification from its multi-segment past. Within that one segment, revenue is disaggregated into two categories — Underwriting and Claims:

(Source: FY2025 10-K, Note 6 disaggregated revenues; MD&A 2025-vs-2024.)

Core products. The Underwriting category (71% of revenue) is the historical heart of the franchise and contains the highest-quality assets:

• Forms, Rules, and Loss Costs — the legacy ISO franchise. Verisk is the U.S. industry-standard author of policy language, prospective loss costs, and rating rules across 32 lines of insurance, processing ~2,000 regulatory filings per year and interfacing with regulators in all 50 states plus four territories. In homeowners alone it maintains ~6 base coverages, 383 national endorsements, and 714 state-specific endorsements (FY2025 10-K, p.5). This is the embedded “operating system” of U.S. P&C underwriting.
• Statistical-agent services — as a licensed/appointed statistical agent in all 50 states, Puerto Rico, and D.C., Verisk aggregates statutorily-required premium and loss data from insurers and reports it to regulators (p.5–6). This is regulatory plumbing few clients can self-perform economically.
• Underwriting Data & Analytics — property/auto rating data on ~143M residential and 16.3M commercial U.S. properties; community fire-protection (PPC) and building-code scoring.
• Extreme Event Solutions (catastrophe modeling) — probabilistic models for hurricanes, earthquakes, wildfire, flood, terrorism, and casualty, across 120+ countries.
• Specialty, Life (FAST platform), and International underwriting software and data round out the category.

The Claims category (29%) comprises Property Estimating Solutions (Xactimate/Xactware — the dominant property-repair-cost estimating platform, with monthly-revised price lists across 468 North American economic areas and ~22,000 unit-cost line items), Anti-Fraud Solutions (the world’s largest P&C claims database — 1.9B claim records, ~193,000 new claims/day), Casualty Solutions (Medicare compliance, workers’-comp reporting, AI medical-record review), and international claims.

Business model — how it makes money. Verisk sells “solutions,” typically through annual subscriptions or one-to-five-year agreements, generally prepaid quarterly or annually and auto-renewed each calendar year. Subscriptions and long-term agreements were over 80% of revenues in 2025 (FY2025 10-K, p.4), with the balance transactional (point-in-time, billed monthly). This produces an unusually attractive financial structure: deferred revenue of $446.2M, ~99% of remaining performance obligations satisfiable within one year, and a structural working-capital deficit — because customers prepay, revenue growth is a source of cash rather than a use (Liquidity section, p.36). Roughly 70% of revenue derives from U.S. P&C primary insurers, ~82% of revenue is U.S.-sourced, and the customer base is diversified (top-50 clients = 42% of revenue; no single client >3%).

Verdict. This is a high-quality, highly recurring revenue model: ~80%+ subscription, prepaid, auto-renewing, embedded in regulatory and operational workflows, with negative working capital and gross margins near 70%. The 2022–2023 divestitures sharpened the story into a single, structurally attractive vertical. The principal quality caveat is the flip side of dominance — with all top-100 U.S. P&C insurers already as clients, U.S. underwriting growth is now overwhelmingly a price-realization story (UW +8.1% organic, almost entirely price), not a unit-growth story — which raises the durability question that the next two sections must answer.

3. Industry Dynamics

Market structure. Verisk operates in the P&C insurance data, analytics, and decision-support market — fragmented at the edges, concentrated at the core. The core “must-have” layers (industry-standard forms/rules/loss costs, statutory statistical reporting, the industry claims-fraud database, property-repair cost estimation, and catastrophe modeling) are each dominated by one or two incumbents, while the periphery (point underwriting data, telematics, niche peril models, insurtech tools) is contested by many. The relevant profit pool sits in the data and model layers, where gross margins run ~70% and incremental copies of a data product cost almost nothing to deliver — the textbook high-fixed-cost / near-zero-marginal-cost economics that, combined with captivity, produce durable scale advantages (Greenwald).

The ISO / advisory-organization role and its regulatory anchoring. Verisk’s structural advantage is inseparable from U.S. insurance regulation. P&C insurance is regulated at the state level; insurers must file rates, rules, and forms with each state’s department of insurance and must report premium/loss statistics. ISO was created in 1971 precisely to perform these collective functions for the industry, and Verisk remains a licensed or appointed statistical agent in all 50 states, Puerto Rico, and D.C. (FY2025 10-K, p.5–6). Two consequences follow. First, advisory-organization status gives ISO’s forms and loss-cost filings a quasi-official standing — regulators expect them, smaller carriers rely on them to demonstrate compliance, and they have been “tested in litigation and tailored to reflect judicial interpretation,” which standardizes claim settlement and reduces litigation cost. Second, the statistical-agent role is a scale monopoly enforced by regulation: it is uneconomic for the industry to support multiple agents performing the same statutory aggregation, so utilization concentrates in one provider — Verisk explicitly notes it “capture[s] significant economies of scale given the level of utilization of this service within the U.S. P&C insurance industry.”

This is the “toll-booth on insurance premiums” archetype: a business whose revenue scales with industry premium and claims volume while bearing none of the underwriting risk. The test that matters is whether the toll is durable — and here the durability comes from dataset uniqueness, switching costs, and regulatory standing rather than from any contractual lock.

Switching costs and barriers to entry. Barriers here are unusually high and stacked: (1) a 50-year contributory-data pool (38.9B underwriting statistical records; 1.9B claims records) that a new entrant cannot replicate at any capital cost, because the data is a function of time and industry trust, not money; (2) regulatory incumbency (statistical-agent licenses in every state; forms tested in case law); (3) deep workflow embedding — ISO forms are the literal policy language carriers issue, and Xactimate is the shared estimating language among carriers, independent adjusters, and contractors, so switching requires re-papering policies or retraining an entire claims supply chain; and (4) data-network effects — the more insurers contribute loss data, the more predictive the loss-cost, fraud, and catastrophe models become, which compels more insurers to subscribe. Contributory-data agreements “remain in effect unless the contributor chooses to opt out,” and “it is rare that contributors elect not to continue providing… all or a substantial portion of their data” (p.11). Much of this data is obtained at no cost to Verisk.

Where regulation distorts the capital cycle (Marathon lens). A normal capital cycle would see high returns (Verisk’s ~45% operating margins) attract capital that builds competing capacity and mean-reverts profitability. That mechanism is largely disabled here because the capacity that matters — a multi-decade contributory database plus statistical-agent appointments — cannot be financed into existence. This is one of Marathon’s explicit “capital-cycle breakdown” conditions (policy/regulatory protection of an incumbent). The corollary, however, is the other breakdown condition — technology disruption — which is the live threat: generative AI and cheaper public data could, over time, erode the value of certain Verisk layers (the 10-K itself flags that “free or relatively inexpensive information” may reduce demand, p.16).

Demand drivers and cyclicality. Demand is tied to U.S. P&C premium volume and policy/claim counts, which are comparatively stable and partly counter-cyclical (catastrophe activity, fraud, and rate-filing activity rise when the environment worsens). Some invoices are linked in part to premiums, giving Verisk a modest organic tailwind from the hard P&C pricing cycle of recent years — but also a downside if premiums soften. The industry is mature and slow-growing in volume; growth comes from price, new analytics layers, international, and life.

Verdict. Structurally good — one of the more attractive analytics niches in the public markets. State-level regulation, a half-century contributory-data monopoly, embedded workflows, and high-fixed-cost economics produce durable barriers to entry and a profit pool capital cannot easily compete away. The honest qualifications: (1) the core U.S. underwriting market is mature and largely penetrated, so industry growth is mid-single-digit and price-led; (2) the same regulatory standing that protects Verisk makes it a target for antitrust and pricing scrutiny (the FTC Second Request that killed AccuLynx is the canary); and (3) the regulatory moat does not protect against the technology-disruption breakdown of the capital cycle — AI is the one force that could re-open competition over the medium term.

4. Competitive Position

Name the moat (Greenwald). Verisk possesses the strongest of Greenwald’s three advantage types — economies of scale combined with customer captivity — further reinforced by government/regulatory incumbency. The mechanism is concrete and tied to financial outcomes:

• Economies of scale via a contributory-data monopoly. Verisk holds the largest U.S. P&C contributory datasets (38.9B underwriting records, 1.9B claims records), assembled over 50 years from the industry itself. Fixed costs of building and maintaining these databases and models are large; the marginal cost of serving one more subscriber is near zero. Crucially — Greenwald’s central condition — scale here is coupled with captivity, so a would-be entrant cannot simply spend its way to equivalent scale: the customers (and their data) are already attached to the incumbent.
• Customer captivity via switching costs and data-network effects. ISO forms are the policy language carriers actually issue and regulators expect; Xactimate is the shared estimating standard across carriers/adjusters/contractors; the claims database’s fraud-detection value rises with every additional contributing insurer. Switching imposes re-papering, retraining, regulatory-refiling, and model-degradation costs on the client — classic demand-side captivity.

The financial proof of the moat. A moat claim must tie to a financial outcome that would deteriorate without it. Here it does: gross margin ~70%, operating margin ~44% (normalized), adjusted-EBITDA margin ~56% in 2025, and — most tellingly — pricing power: 2025 underwriting organic growth of +8.1% was derived primarily from an annual increase in prices, not volume. A business that can raise price mid-to-high-single-digits annually into an already-fully-penetrated client base, year after year, is exhibiting the pricing power that only genuine captivity confers. ROIC of ~36% sits far above cost of capital, satisfying Greenwald’s ROIC test; and market-share stability — the same incumbent holding the statistical-agent role and the standard-forms franchise for decades with no entrant displacing it — satisfies the share-stability test.

Direct competitive comparison (by line). The moat is line-specific, and its strength varies sharply:

The regulatory-incumbency angle cuts both ways. Verisk’s statistical-agent status and standard-forms authority are a moat — but they also mark it as a quasi-utility whose market power regulators police. The clearest recent evidence is the collapse of the $2.35B AccuLynx acquisition: after announcing the roofing-contractor-software deal in July 2025, Verisk received an FTC Second Request on October 22, 2025, failed to clear by the December 26, 2025 deadline, and terminated the deal. A Second Request on a claims-ecosystem adjacency signals that antitrust authorities view Verisk’s position in the claims data/estimating workflow as conferring market power — which simultaneously confirms the moat and caps Verisk’s ability to extend it by acquisition.

The AI/LLM disruption question — the central bear argument. Is the data moat exposed to generative AI? The honest answer is partially, and asymmetrically by layer. (1) The regulatory/statistical-agent and standard-forms franchise is the least exposed — AI does not change the requirement to file regulator-approved language and statutory statistics, nor can an LLM synthesize 50 years of pooled industry loss data that does not exist publicly. (2) The proprietary contributory data is moderately protected — AI is only as good as its inputs, and Verisk owns the inputs; indeed Verisk is deploying generative AI itself (XactAI, AI damage assessment, AI medical-record summarization) to defend and extend. (3) The more commoditizable layers are exposed — point property/auto data, basic estimating, and document-extraction tasks could be replicated by AI plus cheaper public data, exactly the risk the 10-K names. The plausible outcome is margin compression at the edges, not core displacement — but this is the assumption most worth pressure-testing, and it is the live reason the stock trades ~44% below its high.

Verdict. A durable, monopoly-like data moat in its core — among the widest in the analytics universe — but not uniformly impregnable. The forms/rules/loss-cost and statistical-agent franchise is genuinely close to unassailable; Xactimate and the fraud database are strong; catastrophe modeling is a real duopoly with Moody’s RMS; underwriting data/analytics is genuinely contested by LexisNexis and others. Verisk is absent from the large auto-physical-damage claims pool (owned by CCC), faces a bundling threat from core-systems vendors, and confronts a medium-term AI-disruption tail that could erode the commoditizable layers. The moat is real and tied to financial outcomes — it is not a moat that can be assumed to persist unchanged through the AI transition, and the FTC’s posture means it can no longer be widened cheaply by M&A.

5. Growth History and Forward Opportunities

Revenue base and the divestiture reset. Verisk’s reported revenue line is only interpretable once the 2022–2023 portfolio surgery is stripped out. The company sold Financial Services (2022) and Energy/Wood Mackenzie (2023, generating $3,066.4M of disposal proceeds — FY25 10-K cash-flow statement), and now reports a single Insurance segment. On a clean continuing-operations basis the trajectory is: FY21 $2,462.5M → FY22 $2,497.0M → FY23 $2,681.4M → FY24 $2,881.7M → FY25 $3,072.7M — a four-year CAGR of ~5.7%, accelerating to +7.5% in FY24 and +6.6% in FY25. [FACT] (Note: the often-cited FY21 figure of ~$2,998.6M is the total-company number including since-divested units; the apples-to-apples continuing-ops base is $2,462.5M, and using the larger number understates post-divestiture growth.) [INTERPRETATION]

Segment composition and the organic read. FY25 Underwriting was $2,179.9M (+7.7%) and Claims $892.8M (+4.1%); FY24 Underwriting $2,024.3M (+7.0%) and Claims $857.4M (+8.7%) (FY25 10-K, MD&A and segment note). [FACT] Underwriting is structurally the better business — the subscription-heavy, mission-critical industry-standard datasets (ISO forms, statistical-agent data) competitors cannot replicate — while Claims is somewhat more exposed to transactional volume and competition. [INTERPRETATION]

Verisk does not disclose true organic constant-currency (OCC) growth in its 10-K/10-Q — that headline appears only in the earnings exhibits and Investor-Day materials — but the filings report growth “excluding recent acquisitions and disposition,” the best available organic proxy: FY24 +7.2%, FY25 +6.9%, Q1-2026 +5.5%. [FACT] So roughly two-thirds to three-quarters of reported growth is organic, with bolt-ons (Simplitium, SuranceBay, Rocket, Morning Data, Mavera, Krug) adding the rest. That is a high bar of organic content for a data company — growth is earned, not bought. [INTERPRETATION]

Price versus volume, and the subscription engine. The most important quality signal is mix: subscriptions and long-term agreements were over 80% of revenue in 2025, while transactional/advisory work fell to ~17% (from ~19% in 2024). [FACT] The recurring base renews automatically and is largely prepaid, so the cash-conversion cycle is negative (cash precedes revenue recognition). Verisk’s organic growth is overwhelmingly a function of annual price escalators plus modest net-new and cross-sell, not transactional volume — which is why growth has been remarkably stable in a 5–8% band across very different macro environments. The pricing power is real and visible: revenue grew every year while the subscription mix rose, meaning the company is not buying growth with discounting. [INTERPRETATION]

Forward opportunities. Four credible vectors: (1) Extreme Event Solutions (catastrophe modeling), where Verisk is the #2 global player behind Moody’s RMS and climate-driven demand for cat models, secondary-peril modeling, and reinsurance analytics is a secular tailwind; (2) anti-fraud and analytics, where the proprietary contributory databases widen with every customer; (3) international and life expansion (UK $250.1M and “other countries” $300.3M in FY25 — together ~18% of revenue, growing faster than the US base of $2,522.3M); and (4) generative-AI productivity layers marketed to insurers. [FACT/INTERPRETATION] Specific GenAI product names warrant caution — management positioning until they appear in the organic-growth bridge — but the data moat means Verisk can monetize AI features as price uplift on an installed base rather than as speculative new TAM. [INTERPRETATION]

Is growth decelerating? Yes, at the margin. Q1-2026 reported revenue rose just +3.9% ($782.6M vs $753.0M), or +5.5% ex-acquisitions, with Underwriting ex-M&A +6.0% and Claims +4.3% (Q1-2026 10-Q, MD&A). [FACT] That is a step down from the ~6.6–7.5% FY24–FY25 cadence. One quarter is not a trend — currency, transactional timing, and a soft cat-modeling/low-weather renewal cycle can all explain it — but at ~18.5x EV/EBITDA, the market is underwriting durable mid-to-high-single-digit organic growth, and the first-quarter print is the kind of data point that pressures a premium multiple. [INTERPRETATION]

Verdict: high-quality growth, decelerating from a high base. Genuinely high-quality growth — predominantly organic, predominantly recurring, predominantly price-led on a mission-critical subscription base, with a widening data moat. It is not, however, fast growth, and it has slowed to ~5.5% organic in the most recent quarter. The growth is the kind a long-term owner wants (durable, high-margin, low-capital-intensity); the risk is purely that the price paid assumes the deceleration reverses.

6. Financial Quality

Verisk is one of the higher-quality compounders in the data-and-analytics universe, and the financials make the case without adjectives: gross margins near 70%, GAAP operating margins in the mid-40s%, adjusted-EBITDA margins in the mid-50s%, FCF conversion above 1x, and ROIC in the mid-30s%. The complications are (a) two years of one-time items that distort the GAAP trend, and (b) a balance sheet engineered into negative equity by buybacks. Both are explainable; neither is a quality flag once understood.

Income statement and margin trajectory (continuing operations, $M)

* FY22 operating/net margin is distorted by a $354.2M “other operating income” (divestiture-related gain); normalized, FY22 operating income ≈ $1,052M / ~42% margin. † FY24 net is flattered by $95.7M of investment income (vs ~$11–13M normally), a one-time gain. (FY25 & FY23 10-Ks, statements of operations.) [FACT / QoE FLAG]

The clean trend to anchor on: gross margin has climbed 460bps in four years (65.3% → 69.9%) and normalized operating margin has held at ~42–44% while revenue grew ~25%. That is textbook operating leverage on a subscription data business — cost of revenues grew only ~8% cumulatively against ~25% revenue growth. SG&A rose faster (FY23 $391.8M → FY25 $458.2M) but remains ~15% of revenue. [FACT/INTERPRETATION]

Adjusted EBITDA (Verisk’s headline metric)

Verisk reported FY25 adjusted EBITDA of $1,727M (+8.5% OCC), ~56% margin (FY24 ~$1,620M, FY23 ~$1,467M on a like build). [FACT — company-reported FY25; prior years are reconstructed proxies and may differ slightly from Verisk’s official definition.] The mid-50s%-and-rising margin is exceptional even within data/analytics; S&P Global and Moody’s operate in a similar 55–60% range, so Verisk is in the right peer cohort, not an outlier requiring suspicion. [INTERPRETATION]

Cash flow and FCF conversion ($M)

(FY25 & FY23 10-Ks, cash-flow statements.) [FACT/COMPUTED] FCF conversion at or above 1x confirms earnings are cash, not accrual — a function of the prepaid-subscription model. FY25’s 1.31x and the jump in OCF reflect favorable deferred-revenue/working-capital timing as much as underlying growth, so the full ~30% OCF step-up should not be annualized; a normalized run-rate is closer to ~$1.0–1.1B. Capex is low and stable at ~8% of revenue, much of it capitalized software — a genuinely capital-light model. One caution: Q1-2026 OCF fell 12.2% to $390.4M (likely timing, but flagged). [INTERPRETATION/OPEN QUESTION]

Share count, buybacks and dilution

(SEC EDGAR XBRL; cash-flow statements.) [FACT] Diluted shares fell ~15% from FY20 to FY25, and a further $1.5B ASR closed in February 2026, driving the count toward ~131M. Critically, SBC is only ~1.7% of revenue and flat, far below software peers, so buybacks shrink the net share count rather than merely offsetting dilution — real per-share value creation, not optical churn. [INTERPRETATION]

Balance sheet, leverage and the negative-equity mechanics

FY25 (12/31/25): total debt $4,737.2M ($1,508.9M short-term + $3,228.3M long-term) against cash of $2,178.2M → net debt ~$2,559M. [FACT] (Correcting an earlier baseline: 12/31/25 cash was $2,178M, not ~$525M — Verisk pre-funded the February-2026 $1.5B ASR, drawing $1.5B of short-term debt and parking the cash, which is why both cash and short-term debt spiked at year-end.) On adjusted EBITDA, net debt/EBITDA ≈ 1.5x (gross ~2.7x) and interest coverage ≈ 7.9x — solidly investment-grade. [COMPUTED]

The negative-equity question. Stockholders’ equity was a small positive $309.0M at 12/31/25, then turned to a deficit of −$1,167.7M at 3/31/26. [FACT] The mechanics: retained earnings of $7,810.5M are more than offset by treasury stock carried at cost of −$10,721.8M (rising to −$12,132.0M after the Feb-2026 ASR). In plain terms, Verisk has spent years buying back stock at prices vastly above book; under GAAP the full repurchase cost lands in the treasury-stock contra-equity line, mechanically driving book equity negative. This is an artifact of aggressive (and largely value-accretive) buybacks, not financial distress — the company generates ~$1.2B of FCF and carries IG leverage. The practical implication: ROE is meaningless (negative denominator) and must be ignored; the right lens is ROIC. [INTERPRETATION]

Returns on capital. NOPAT ≈ operating income × (1 − ~22.4% tax) ≈ $1,043M over invested capital (net debt $2,559M + equity ~$310M) ≈ $2,869M ⇒ ROIC ≈ 36% for FY25; even on a grosser invested-capital base it remains comfortably above 20%. [COMPUTED — ASSUMPTION: eff. tax 22.4%] Returns of this magnitude on a capital-light, recurring-revenue base are the financial signature of a real moat. Goodwill + intangibles of $2,224.8M (~36% of assets) is moderate for a serial acquirer and not a red flag; no recent impairments. [FACT]

Quality-of-earnings flags (summary). (1) FY22 $354.2M divestiture gain and FY24 $95.7M investment gain distort GAAP operating/net margins — normalize before any trend claim. (2) FY21–FY22 OCF figures are total-company (include discontinued ops) and are not clean continuing-ops comparables. (3) FY25 OCF benefited from working-capital timing; Q1-2026 OCF dipped 12.2%. None undermine the core picture — they require care, not skepticism about the franchise.

Verdict: economics clearly improve with scale. Gross margin up 460bps in four years, normalized operating margin holding in the low-40s%, adjusted-EBITDA margin in the mid-50s% and rising, ~1x+ FCF conversion, ~8%-of-revenue capex, low/flat SBC, and ~36% ROIC together describe a business with genuine operating leverage and a capital-light, cash-generative model. The negative book equity is a buyback artifact, not a weakness. The only blemishes are GAAP one-time items (cosmetic) and the question of whether mid-single-digit organic growth justifies the leverage now applied to the equity — a debate that belongs in Valuation, not here.

7. Capital Allocation

Verisk’s capital-allocation story since 2022 is dominated by one decision: management dismantled a three-segment conglomerate into a pure-play insurance-data company and returned the proceeds to shareholders at scale. The execution of the strategy was clean; the price discipline of the capital return was not.

The refocus and the proceeds. Verisk sold Financial Services to TransUnion for ~$515M (2022) and Energy/Wood Mackenzie to Veritas Capital for $3.1B cash plus up to $200M contingent (announced 10/31/2022, closed Q1 2023; a $131.1M loss on sale was booked). The combined ~$3.6B of proceeds, plus balance-sheet capacity, funded a very large capital return. The Board authorized a $3.0B repurchase effective 2/1/2023, executed mostly through a $2.5B accelerated share repurchase (ASR) in March 2023 (initial delivery 10,655,301 shares at a $187.70 reference; total FY2023 treasury purchases 12,849,921 shares for ~$2,801.2M). [FACT] This is the single most consequential capital act of the period — and, in hindsight, well-timed: the 2023 ASR struck near $188, essentially today’s price.

Buyback cadence and prices. After 2023 the program decelerated as proceeds were exhausted: repurchase spend was $2,762.3M (FY23), $1,005.0M (FY24), $624.0M (FY25). The FY24–FY25 buybacks ran through smaller ASRs — Nov 2024 $300M, Mar 2025 $200M, May 2025 $100M, Aug 2025 $50M — plus a Q4 2025 enhanced open-market program ($223.8M at avg $218.99). The blunt fact: the ASR final-settlement prices were $278.92 (Nov’24), $288.09 (Mar’25), $309.58 (May’25), $263.66 (Aug’25). [FACT] Against ~$182 today, essentially the entire ~$1.6B repurchased in FY24–FY25 was bought 20–70% above current levels, much of it near the all-time-high zone ($290–$310). [INTERPRETATION] Verisk’s repurchase machine is calendar- and proceeds-driven, not valuation-driven — ASRs deliver predictable mechanical share reduction, and the firm leaned on them precisely when the stock was richest. Authorization remains ample (~$967.5M at 12/31/2025; Board topped the total to $2.5B). The forward question is whether management buys more aggressively now that the stock is down ~44% — and whether the pace deceleration reflects exhausted firepower or restraint.

Share count and dilution. Diluted weighted shares fell from 163.3M (FY21) to 140.1M (FY25), ~14% — almost entirely the 2023 mega-buyback. SBC is modest and well-controlled (~1.7% of revenue), and buybacks have more than absorbed grant dilution — a clean profile and a point in management’s favor.

Dividends. Small but reliable: cash paid $196.8M / $221.7M / $251.2M (FY23/24/25), quarterly raised to $0.45 in 2025 (~$1.80 annualized). Payout is conservative at ~28% of FY2025 net income; yield ~1.0–1.1%. The dividend signals confidence and grows steadily, but buybacks are the primary return vehicle.

M&A discipline. Post-refocus M&A has been small, on-strategy insurance bolt-ons: SuranceBay ($163.1M, 7/2025), Krug ($43.3M), Mavera ($28.3M), Simplitium ($19.7M), Rocket ($10.1M) — vertical-adjacent and consistent with the data/analytics narrative. The exception is AccuLynx: a $2.35B all-cash deal agreed 7/29/2025 for a roofing/contractor-workflow SaaS platform — Verisk’s largest post-refocus deal and a step outside the core vertical — terminated 12/26/2025 after the FTC failed to clear it. The $1.5B of senior notes raised to fund it triggered a Special Mandatory Redemption (redeemed in full) and the bridge loan was unwound. [INTERPRETATION] Read two ways: an execution/regulatory setback (a year of attention and financing cost for nothing), but also a bullet dodged — paying $2.35B for a contractor-SaaS asset afield from the moat was avoided, and management unwound the financing cleanly rather than overpaying into a forced deal.

Incentive alignment (proxy, DEF 14A 4/3/2026). The short-term cash incentive pays on Revenue and Adjusted EBITDA (plus a 20% individual modifier); FY2025 landed at Revenue $3,024M = 95% and Adjusted EBITDA $1,699M = 105%, aggregate 100% of target. Long-term equity mix is 40% Relative-TSR PSUs (vs S&P 500), 20% ROIC PSUs (3-yr avg ROIC), 20% restricted stock, 20% options. CEO Lee Shavel’s FY2025 total comp was $13,540,847 (pay ratio 196:1). Say-on-pay passed ~95.7% (5/19/2026). [FACT] The design is above-average — heavy reliance on relative TSR and a hard ROIC gate ties pay to capital efficiency and to beating the market, exactly what you want at a serial repurchaser. The blemish: STI runs on absolute (not organic) Revenue/EBITDA and paid full target in a year the stock fell ~18% (1-yr TSR −18.1%) — the classic short-vs-long-term tension the LTI is meant to, and partly does, correct. [INTERPRETATION]

Verdict: mixed, leaning competent-but-not-shrewd. The strategic surgery was right — shedding two non-core businesses to become a focused, asset-light, high-ROIC compounder, returning proceeds promptly rather than empire-building. Dilution control, dividend discipline, and incentive design (ROIC + relative TSR) are genuinely good. But the repurchase program — the dominant use of cash — was executed without price discipline: ~$1.6B spent in FY24–FY25 at $264–$310 looks poorly spent against $182, and the pace then slowed as the stock cheapened. AccuLynx is a wash-to-small-positive. This is a team that allocates capital systematically and cleanly but not opportunistically — fine for a steady compounder, a missed opportunity for value creation through the cycle.

8. Changes and Headwinds — Last Two Years

Strategic transformation completed. The defining change is the completion of the multi-year shift from diversified-analytics conglomerate to focused insurance-data pure-play. With Financial Services (2022) and Energy/Wood Mackenzie (Q1 2023) gone, FY2024–FY2025 is the first clean run-rate of “new Verisk”: a single Insurance segment, ~$3.07B revenue, ~56% adjusted-EBITDA margin. This simplifies the story and concentrates the business on its strongest moat, but removes diversification — the company now rises and falls with one end market.

Leadership. Lee Shavel (former CFO) is CEO, completing the handoff from long-tenured CEO Scott Stephenson; Elizabeth Mann is CFO. The C-suite is the team that ran the refocus — continuity for the strategy and capital-return framework.

Investor Day and the long-term framework. At its March 2026 Investor Day, Verisk reiterated its medium-term framework: OCC revenue +6–8%, OCC adjusted-EBITDA +7–10%, and adjusted-EBITDA margin expansion of 25–75 bps per year. [FACT] FY2025 delivered inside that frame despite a headwind — Revenue $3,073M (+6.6% OCC) and Adjusted EBITDA $1,727M (+8.5% OCC) — achieved against an unusually low level of weather/catastrophe activity that suppressed transactional revenue. FY2026 guidance is revenue $3.19–3.24B and adjusted EBITDA $1.79–1.83B. Net income was $908.3M (vs $958.2M FY2024) and diluted EPS $6.48 (down from $6.71), the EPS dip reflecting the absence of FY2024 one-timers and a smaller share-count benefit as buybacks slowed. [INTERPRETATION] The model is performing on plan; the reiterated targets and an in-range year of mid-single-digit OCC growth with margin expansion are the bedrock of the bull case — but unspectacular for a stock that traded at ~40x earnings at its peak.

The AccuLynx episode. Agreed 7/29/2025 ($2.35B cash), terminated 12/26/2025 after the FTC declined to clear it, triggering the Special Mandatory Redemption of the $1.5B senior notes and the bridge unwind; a new Term Credit Agreement (2/23/2026) managed post-unwind liquidity. The most material discrete event of the two years — a large, off-vertical deal the regulator effectively blocked.

Governance shifts. Shareholders adopted a 25%-threshold special-meeting right, and at the 5/19/2026 annual meeting a shareholder proposal for the right to act by written consent PASSED against the board’s recommendation (59.4M For / 52.1M Against). A modest but real signal of shareholder assertiveness — the board lost a governance vote — though neither change alters control dynamics materially.

The share-price collapse. The stock is ~$182 vs a ~$323 high (−44%). This is a multiple de-rating, not a fundamental break — FY2025 grew OCC mid-single-digits and guidance steps up — driven by the unwind of a peak SaaS-like multiple, rising-rate pressure on long-duration compounders, the AccuLynx distraction, and a soft-weather year that made the top line look pedestrian. That divergence (steady fundamentals, collapsing price) is the central tension of the thesis.

Verdict: net neutral-to-slightly-strengthening on fundamentals, weakened on sentiment/optionality. The completed refocus and an on-plan, reiterated framework strengthen the core. The AccuLynx failure weakens it at the margins (regulatory limits on inorganic growth; a flicker of strategic drift; a year of focus lost). The governance loss is a minor negative. The 44% drawdown is the most important change — it does not damage the business but transforms the setup, converting a perennially-expensive quality name into a far more reasonably-priced one.

9. Risk Analysis

The dominant risks are not balance-sheet or solvency risks — leverage is investment-grade and cash generation is robust — but growth-durability and disruption risks against a still-premium multiple. The matrix below synthesizes the analysis.

Catastrophic-loss / total-loss assessment. The probability of a permanent capital impairment is low. Verisk is not a leveraged or cyclical balance sheet; the realistic adverse scenario is multiple compression plus growth disappointment (a de-rating toward the commoditized-data peer band), not insolvency. The genuine tail risk is a slow structural erosion of the data moat by AI/regulation over many years — a value-decay scenario, not a sudden loss.

10. Valuation Discussion

This section contains no price target and no recommendation. It frames valuation as embedded expectations and as a scenario set; implied value ranges are the analytical output of stated assumptions, not advice.

Multiples and peer comparison

Verisk trades at ~$181.73, ~44% below its 52-week high. On reconciled FY25 figures: trailing P/E ~27.7x, forward P/E ~21x, EV/EBITDA ~18.5x, EV/Revenue ~9.0x, P/S ~7.7x, ~1.1% dividend yield, ~28% payout. Using post-ASR net debt of ~$4.0B, EV reconciles to ~$27.8–27.9B. The correct comparison set is information-services / fee-based franchises, not P&C carriers — Verisk sells data into the insurance value chain and bears no underwriting risk, the same “toll-booth” economics that justify 25–30x P/E for fee-based insurance services vs 11–14x for risk-bearing carriers.

(Public market-data aggregators, 2026-06-06; sanity-checked. Growth is TTM, not strictly organic.)

Two readings stand out. First, the de-rating is largely sector-wide, not Verisk-specific. The entire regulated-data/analytics complex re-rated lower from 2024 peaks on AI-disruption fears — SPGI, MCO, and FDS all compressed (FDS now ~13x forward). Verisk’s ~21x forward P/E sits between its closest structural analogs SPGI (~19x) and MCO (~24x) and at a clear premium to commoditized credit/data names (TRU ~13x, EFX ~17x, IT ~11x). On EV/EBITDA, VRSK (18.5x) is essentially on top of SPGI and below MCO/MSCI. So at the multiple level Verisk is priced like a high-quality data monopoly, in line with its nearest peers. Second, Verisk is cheap against its own history — 19th percentile on P/E, 25th on P/S, 22nd composite over ~10 years. The caveat: Verisk also has the slowest reported growth in the table, which partly justifies a discount to faster peers (MSCI, EFX) and explains a sliver of VRSK-specific overhang atop the sector move.

FCF-yield and cash-return lens

FY25 FCF of $1,191.9M (~38.8% margin, ~1.3x net income — flattered by working-capital timing, so a normalized base is ~$1.0–1.1B) puts the FCF yield at ~5.0% on market cap and ~4.3% on EV. Layered with a ~1.1% dividend (28% payout) and the ~$1.5B ASR (~6% of cap), total shareholder yield is mid-to-high single digits, internally funded at ~1.5x net leverage. For a ~36%-ROIC, capital-light business, a ~5% FCF yield is a defensible entry yield if the franchise grows; unremarkable if growth stalls.

Reverse-DCF / embedded expectations

At ~$181.73 and EV ~$27.9B, a simple reverse-DCF (WACC 8.5%, terminal growth 3%, normalized FCF base ~$1.15B) implies the market is underwriting ~5.5–6.5% steady-state FCF growth — the low end of management’s 6–8% framework. [ASSUMPTION on WACC/terminal; INTERPRETATION on implied growth.] The de-rating from a ~33x peak forward multiple to ~21x has compressed embedded expectations to below what management says the franchise can deliver. The bear’s burden is to argue even ~5–6% is too optimistic (AI + saturation drag growth toward low-single-digits); the bull’s burden is simply that the 6–8% framework holds and the multiple normalizes.

Scenario analysis (bear / base / bull)

Three-to-five-year scenarios; share count ~131M post-ASR with modest continued buyback. These ranges are the arithmetic output of the stated assumptions — not a recommendation or target.

(All inputs explicit; zones derived from forward EPS off the FY25 $6.48 base compounded at the scenario growth × exit multiple, net of buyback. Bull approaches the prior 52-wk high.) The asymmetry: the current price (~$182) sits at the low edge of the Base zone — the market is roughly pricing the Base case’s pessimistic edge. Downside to Bear is ~12–26%; upside to mid-Base ~10–20% and to Bull substantial — but the Bull requires a re-acceleration the Q1-26 organic print (+5.5%) does not yet support.

What the market is pricing correctly vs incorrectly

Likely correct: that Verisk deserves an information-services multiple, not a carrier multiple — ~18.5x EV/EBITDA is appropriately in the SPGI band and above commoditized peers, consistent with the durable moat and ~36% ROIC; and that the 33x peak should not be extrapolated. Possibly incorrect (the debate): the market has compressed embedded growth to the low end of the framework and is treating the deceleration (FY24 +7.2% → Q1-26 +5.5%) as a durable step-down rather than a tough-comp/low-weather trough in transactional lines. If the deceleration is cyclical and the 6–8% framework re-asserts, today’s price under-rates the franchise; if AI/saturation makes ~5.5% the new ceiling — or erodes pricing power in the core data products — then even ~21x is not cheap and the Bear zone governs. The valuation does not resolve that question; it shows the price is set near the dividing line.

11. Variant Perception

Consensus belief. The sell-side consensus rates Verisk roughly Hold-to-modest-Buy (rating ~3.4/5; third-party mean target ~$220 — cited as context only). Consensus accepts that Verisk is a high-quality, wide-moat data compounder, but the price action (−44%) reveals the operative market view: that the growth algorithm is resetting lower and the long-duration premium multiple is no longer warranted in a higher-rate, AI-uncertain world. The market is, in effect, treating Verisk as a “great business that finally got de-rated to merely-fair,” with residual fear that fair could become expensive if growth keeps slowing.

The strongest bull case. Verisk is a regulated-data monopoly with ~36% ROIC, >80% recurring prepaid revenue, durable annual pricing power, and a 50-year contributory-data asset no one can replicate — now available at the cheapest quintile of its own decade (~21x forward) after a sector-wide de-rating that conflated it with more commoditized data names. The organic deceleration is cyclical (tough comps, an unusually low-weather/catastrophe year that suppressed transactional revenue), not structural; the 6–8% framework, reiterated at the March 2026 Investor Day and delivered in FY25 (+6.6% OCC revenue, +8.5% OCC EBITDA), re-asserts. AI is a tailwind Verisk monetizes on its installed base (XactAI, AI claims/medical review) rather than a threat, because models are only as good as the proprietary inputs Verisk owns. Steady mid-single-digit revenue + margin expansion + buyback drives high-single-digit EPS growth, and a normalization of the multiple toward the SPGI band delivers the Base/Bull zones.

The strongest bear case. Verisk is a structurally mature, single-vertical business whose core U.S. underwriting market is saturated (all top-100 carriers already clients), so growth is almost entirely price — and annual price escalation into a stressed carrier base is precisely the kind of toll that eventually meets resistance. Organic growth is on a visible downtrend (toward +5.5%), the multiple is still a premium ~21x, and generative AI plus cheaper public data is a real solvent for the commoditizable layers (point data, basic estimating) over the medium term. The FTC’s blocking of AccuLynx shows the moat can no longer be widened by M&A and that regulators are wary of Verisk’s market power — a cap on both growth optionality and, potentially, core pricing. Management compounded the setup by spending ~$1.6B on buybacks at $264–$310. If ~5% becomes the ceiling, the stock de-rates toward the commoditized-data band (Bear zone).

The 3–5 assumptions that matter most:

1. Is annual price escalation in the core ISO/underwriting franchise durable? (The load-bearing wall — pricing power is the moat’s financial expression.)
2. Is the organic deceleration to +5.5% cyclical or structural? (Determines whether ~21x is cheap or fair.)
3. Does generative AI erode, preserve, or extend the value of Verisk’s proprietary data layers?
4. Will management redeploy capital opportunistically now that the stock is cheap (buy back harder at $182), or keep mechanically ASR-ing?
5. Does regulatory/antitrust pressure stay confined to M&A, or eventually reach core pricing/data-dominance?

What would falsify each side. Bull falsified: two-plus quarters of sub-5% organic growth with explicit evidence of carrier pricing pushback or accelerating share loss in contested lines (LexisNexis/CCC). Bear falsified: organic growth re-accelerating back through ~7% on renewed pricing + cross-sell, with AI products showing up as a positive contributor in the organic-growth bridge.

12. Fact vs. Interpretation Table

13. Open Questions

1. Hard client-retention rate is not disclosed in filings (high-90s widely cited but unverified; proxies: all top-100 clients, auto-renewal, ~99% POs <1yr, rare data opt-out). [OPEN]
2. Is the Q1-2026 +5.5% organic print cyclical (tough comps, low weather, transactional timing) or a structural step-down? The single pivotal question for valuation. [OPEN]
3. Exact post-ASR cash/net-debt at 3/31/26 (the year-end figure was distorted by ASR pre-funding). [OPEN]
4. Official adjusted-EBITDA reconciliation vs the reconstructed proxy for FY23–FY24 (FY25 $1,727M is company-reported). [OPEN]
5. Pace and price of forward buybacks — will management deploy harder now the stock is ~44% off its high, or keep mechanically ASR-ing? [OPEN]
6. Competitive share trajectory in contested lines (underwriting data/analytics vs LexisNexis; auto claims, where Verisk is largely absent vs CCC). [OPEN]
7. Q1-2026 OCF −12.2% — working-capital timing or early sign of conversion deterioration? [OPEN]

14. What Must Be True (with falsification tests)

For the BULL case to be correct:

• Durable pricing power — Verisk must continue to raise prices mid-to-high-single-digits annually in the core underwriting franchise without triggering carrier pushback or accelerated in-sourcing. Falsification: two-plus quarters of underwriting organic growth below ~5% accompanied by evidence of price resistance or churn.
• Cyclical, not structural, deceleration — the slowdown toward +5.5% must reverse as comps normalize and weather/catastrophe activity reverts. Falsification: organic growth fails to re-accelerate above ~6% over the next 12–18 months despite a normal-weather year.
• AI is a tailwind, not a solvent — Verisk’s proprietary data must keep AI competitors at bay, and Verisk’s own AI products must show up as positive contributors. Falsification: a credible AI/insurtech entrant or carrier-internal model demonstrably wins share in a core Verisk line, or AI-driven commoditization shows up as pricing pressure.

For the BEAR case to be correct:

• ~5% is the new growth ceiling — saturation and AI cap organic growth in low-single-digits structurally. Falsification: a return to ~7% organic growth on renewed pricing + cross-sell.
• Multiple de-rates further toward the commoditized-data peer band (~13–15x) as the market loses faith in the algorithm. Falsification: the multiple stabilizes/normalizes toward the SPGI band (~18–20x) as growth holds.
• Pricing power cracks or regulation reaches core pricing — the toll meets resistance, whether from stressed carriers or antitrust scrutiny of data dominance. Falsification: continued full price realization and no regulatory action beyond the AccuLynx M&A block.

15. Source Appendix

The full, dated source list — primary SEC filings (FY2023/24/25 10-Ks, the FY2026-Q1 10-Q, 8-Ks, DEF 14A 2024–2026), EDGAR XBRL facts, public market-data, and external trade/industry sources — is collected in Appendix B below. Every non-obvious fact in this memo traces to a primary or public source listed there.

Appendix A — Diligence Questionnaire

Verisk Analytics, Inc. (NASDAQ: VRSK) — Diligence Questionnaire

Supplemental to the research memo. Answers are grounded in the underlying primary sources; Fact/Interpretation/Assumption labels applied where material. The Greenwald (Competition Demystified) and Marathon (Capital Returns) frameworks are applied where they add insight.

General

What thoughtful questions have other investors asked about this company? The recurring institutional questions cluster around four themes: (1) Is the organic growth deceleration cyclical or structural? — i.e., is +5.5% Q1-2026 a tough-comp/low-weather trough or a new ceiling. (2) Does generative AI erode the data moat, or does Verisk monetize it? — the question that drove the 44% de-rating. (3) Is pricing power durable given the core US underwriting market is fully penetrated and growth is almost entirely price, not units. (4) Will management redeploy capital opportunistically now that the stock is ~44% off its high, or keep mechanically buying via ASRs (it bought ~$1.6B at $264–$310 in FY24–25). A fifth, post-AccuLynx: how much growth optionality is lost if the FTC effectively bars sizeable M&A?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? [INTERPRETATION] Neither extreme. FY2025 was achieved against an unusually low level of weather/catastrophe activity that suppressed transactional revenue, so the transactional component is arguably cyclically depressed; the subscription base (>80%) is structurally stable and not cyclical. Margins are near a structural high (adj-EBITDA ~56%), so margin tailwind from here is limited (management guides +25–75 bps/yr).

Driven by external environment or internal actions? Predominantly internal — annual contractual price escalators and cross-sell on a recurring base. The external lever (P&C premium volume, catastrophe activity) modulates the transactional ~17% and some premium-linked invoices, but does not drive the core.

How stable are revenues? Very. >80% recurring, prepaid, auto-renewing subscriptions; ~99% of remaining performance obligations satisfiable within a year; rare contributory-data opt-outs; no single client >3% of revenue, top-50 = 42%. Organic growth has held in a 5–8% band across very different macro environments. [FACT, FY2025 10-K]

Outlook for products/services? Mid-single-digit OCC revenue growth (management framework 6–8%), led by Underwriting (forms/loss costs, cat modeling, anti-fraud) with Claims (Xactimate, casualty, anti-fraud) slower. Growth vectors: catastrophe/climate modeling, international/life, AI productivity layers.

How big is this market — growing, shrinking, domestic or international? The US P&C insurance data/analytics market is mature and slow-growing in volume (premiums grow low-to-mid-single-digits); Verisk’s growth is price + new analytics layers + international/life expansion. ~82% of revenue is US-sourced; international (UK ~8%, other ~10%) grows faster off a small base. [FACT]

Business Quality & Competitive Moat

Is the industry getting more or less competitive? [INTERPRETATION] Bifurcating. The regulated core (forms/loss costs/statistical agent) is not meaningfully more competitive — it is regulation- and data-protected. The edges (point underwriting data, basic estimating, document extraction) are getting more competitive as generative AI lowers entry costs and LexisNexis/CCC/insurtechs press. Net: stable-to-slightly-more-competitive at the periphery, stable at the core.

How profitable is the business (ROIC, ROE)? ROIC ~36% (FY2025, computed NOPAT/invested capital). ROE is not meaningful (negative book equity from cumulative buybacks). Gross ~70%, normalized operating ~44%, adj-EBITDA ~56%, net ~29%. [FACT/COMPUTED] These are top-decile economics.

How profitable is the industry — how many competitors, what barriers to entry? The core data layers are oligopoly/monopoly profit pools (Verisk, plus Moody’s RMS in cat, LexisNexis/RELX in UW data, CCC in auto claims). Barriers are very high: 50-year contributory data (38.9B UW + 1.9B claims records), regulatory incumbency (statistical agent in all 50 states), and embedded workflows. Marathon lens: the capital cycle is “broken” in the incumbent’s favor by regulation/data — capacity cannot be financed into existence.

Can the business be easily understood? Yes — a data/analytics subscription toll on the P&C insurance industry. The only complications are GAAP one-time items (divestiture/investment gains) and the buyback-driven negative equity, both explainable.

Can it be undermined by foreign low-cost labor? No — the moat is proprietary data + regulatory standing, not labor cost. Offshoring affects only the cost base at the margin.

Do brands matter? Yes, in a B2B sense: “ISO” and “Xactimate” are industry-standard names that function as quasi-regulatory and workflow defaults; they carry trust and switching-cost weight rather than consumer-brand value.

What is the nature of competition? Line-specific. Near-monopoly in forms/loss costs/statistical agent; duopoly with Moody’s RMS in catastrophe modeling; contested with LexisNexis/CoreLogic/TransUnion in underwriting data; dominant via Xactimate in property estimating; largely absent in auto physical-damage claims (CCC dominates). The competitive frontier is AI-enabled substitution at the edges.

Customers’ switching costs? High: ISO forms are the actual policy language carriers issue and regulators expect; Xactimate is the shared estimating language across carriers/adjusters/contractors; switching imposes re-papering, retraining, regulatory refiling, and model-degradation costs. Data-network effects compound captivity.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Yes — the most valuable asset, the 50-year contributory database (much of it acquired at no cost), is not capitalized at anything near economic value. This understates “true” book capital and is a reason book-equity-based metrics (ROE, P/B) are meaningless here. [INTERPRETATION]

Off-balance-sheet liabilities? None material flagged; standard operating leases. Contingent consideration on small acquisitions is immaterial.

How conservative is the accounting? Reasonably conservative on the core (subscription revenue recognized ratably; capex/capitalized software ~8% of revenue, disclosed). Caution flags: FY22 $354.2M divestiture gain and FY24 $95.7M investment gain inflate GAAP trend lines — normalize before comparing. Adjusted EBITDA is a management metric; reconcile to GAAP. No revenue-recognition aggressiveness identified. [QoE]

How CapEx-hungry is the business? Light — ~8% of revenue (~$224–244M), largely capitalized software. The model is asset-light; growth funds itself via negative working capital (prepaid subscriptions).

Capital Allocation & Management

How much FCF does the business generate; how is it used; what is the philosophy? FY25 FCF ~$1,192M (normalized ~$1.0–1.1B). Philosophy: return the bulk to shareholders — primarily buybacks (mechanical ASRs), secondarily a small, growing dividend (~28% payout), with small on-strategy bolt-on M&A. [FACT]

Significant acquisitions recently? Small bolt-ons (SuranceBay $163M, Krug, Mavera, Simplitium, Rocket). The one large attempt — AccuLynx ($2.35B, 7/2025) — was terminated 12/2025 after the FTC failed to clear it. [FACT]

Buying back shares? Yes, aggressively — $2,762M (FY23), $1,005M (FY24), $624M (FY25), plus a $1.5B ASR closing Feb-2026. Diluted shares fell ~15% (165M→140M, →~131M post-ASR). Critique: FY24–25 buybacks were executed at $264–$310 vs ~$182 now — poor price discipline. The 2023 ASR (~$188) was well-timed. [FACT/INTERPRETATION]

Issuing large amounts of new shares to insiders? No — SBC is modest (~1.7% of revenue) and flat; buybacks more than offset dilution.

Compensation policy of directors/management? STI on absolute Revenue + Adjusted EBITDA (FY25 paid 100% of target). LTI: 40% relative-TSR PSUs, 20% ROIC PSUs, 20% RSUs, 20% options — above-average alignment to capital efficiency and beating the market. CEO Lee Shavel FY25 comp $13.5M. Say-on-pay ~95.7%. Blemish: STI paid full target in a year shareholders lost ~18%. [FACT]

Motivations of management? [INTERPRETATION] Management ran the refocus and is incentivized on ROIC and relative TSR — appropriately long-term and capital-efficiency-oriented. The buyback indiscipline suggests a systematic/mechanical (not opportunistic/contrarian) capital temperament. A cluster of director open-market buys at ~$180 (Feb 2026) signals board-level confidence; no officer open-market buying.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No — a US C-corp common stock on NASDAQ; standard 1099 treatment. [FACT]

Dividend policy? Quarterly cash dividend, $0.45/qtr (~$1.80 annualized), ~1.1% yield, ~28% payout — growing, but secondary to buybacks. [FACT]

How profitable is the business? Top-decile: ~36% ROIC, ~56% adj-EBITDA margin, >1x FCF conversion (see above).

Is net income diverging from cash from operations? Cash generation is stronger than net income (FCF/NI ~1.3x FY25), confirming high earnings quality — though FY25 OCF was timing-flattered and Q1-26 OCF dipped 12.2% (watch item). No adverse accruals divergence. [FACT/OPEN QUESTION]

Risks & Downside

What factors would cause the stock to decline? Further organic deceleration (sub-5%), evidence of pricing pushback in the core franchise, AI/insurtech share loss in contested lines, a further multiple de-rating toward commoditized-data peers, or regulatory action reaching core pricing. A soft P&C premium cycle would trim the premium-linked/transactional revenue.

Risk of a catastrophic loss? [INTERPRETATION] Low. Investment-grade balance sheet (~1.5x net leverage, ~8x coverage), ~$1.2B FCF, no solvency or cyclical-collapse risk. The negative book equity is mechanical, not distress.

Chance of a total loss? Negligible in any foreseeable scenario. The realistic adverse case is value decay (multiple compression + slow moat erosion by AI/regulation over many years), not permanent capital impairment from insolvency.

Appendix B — Source Appendix

Verisk Analytics, Inc. (NASDAQ: VRSK) — Source Appendix

Public sources used in the research memo and diligence questionnaire, prioritized primary-first. Access date for all quantitative/web pulls: 2026-06-06 unless noted.

1. Primary — SEC filings (EDGAR, CIK 0001442145)

2. Primary — EDGAR XBRL companyfacts / companyconcept

• us-gaap:RevenueFromContractWithCustomerExcludingAssessedTax — multi-year revenue (FY16–FY25).
• PaymentsForRepurchaseOfCommonStock, share count, cost-of-revenue, operating income, cash-flow line items — reconciled to the 10-K statements.
• Form 4 corpus (insider transactions, CIK 1442145) — 154+ filings since 2024-01-01; transaction-code tally and director open-market-buy cluster (Feb 2026).

3. Quantitative helpers (reconciled to filings; unofficial)

• Public market-data aggregators (quote, stats, comps) — price $181.73, market cap ~$23.8B, EV ~$27.9B, multiples (fwd P/E ~21x, EV/EBITDA ~18.5x), and the peer-comp table (SPGI, MCO, MSCI, FDS, TRU, EFX, IT). Accessed 2026-06-06. Treated as convenience data, reconciled to the 10-K.
• Own-history valuation percentiles (P/E 19th, P/S 25th, composite 22nd, ~10y), snapshot (sector, employees, short interest 6.5% of float). Primary EDGAR data used for statement-level figures.

4. External — industry, regulatory, trade press

• Insurance Journal — “Verisk Ends Effort to Acquire AccuLynx” / termination coverage (Dec 2025). https://www.insurancejournal.com/news/national/2025/12/30/852657.htm
• Verisk newsroom — “Verisk Ends Effort to Acquire AccuLynx”; March 2026 Investor Day “Verisk Reiterates its Growth Targets…”. https://www.verisk.com/company/newsroom/
• CCC Intelligent Solutions — AI / auto-claims technology (competitive context). https://www.cccis.com/our-technology/ai
• Risk & Insurance — “How underwriting and claims are reshaped by AI in insurance.” https://riskandinsurance.com/how-underwriting-and-claims-are-reshaped-by-ai-in-insurance-and-how-they-stay-the-same/
• GlobeNewswire — Wood Mackenzie / Veritas Capital divestiture (10/31/2022); FY2025 results release (2/18/2026).

5. Frameworks

• Greenwald & Kahn, Competition Demystified (barriers-to-entry / advantage taxonomy; ROIC & share-stability tests); Chancellor/Marathon, Capital Returns (supply-side capital-cycle analysis; capital-cycle “breakdown” conditions — regulation and technology disruption).