WTW Survey Quantifies the Analytics Dividend: P&C Leaders Hold a 6-Point Combined Ratio Advantage

Fifty-nine North American P&C carriers participated in WTW's March 2026 Advanced Analytics and AI Survey, reporting from senior positions in analytics, actuarial, and strategy functions, and the performance data they provided across the 2022 to 2024 measurement window is precise enough to serve as a benchmark rather than a directional signal: carriers at the leading edge of analytics adoption ran combined ratios six percentage points below slower-adopting peers while growing premium three percentage points faster over the same period. Both numbers matter. The combined ratio gap reflects underwriting quality, claims management efficiency, and risk selection accuracy operating together. The premium growth differential is more telling: it means analytics leaders were not simply protecting margin by being selective, they were confident enough to grow while others managed defensively. In a period that included post-pandemic reserve development, elevated severity inflation, and back-to-back catastrophe years, that combination is not coincidence.

The survey, published March 19, 2026, covers commercial lines, personal lines, and specialty carriers operating in the United States and Canada. This article examines what the WTW data reveals about the anatomy of the six-point gap, where the largest untapped analytics ROI sits inside most P&C carriers, what separates the carriers achieving these results from the majority still at earlier stages of adoption, and what the compounding trajectory looks like through 2030 using Morgan Stanley's expense ratio projections alongside the WTW performance record.

The 59-Carrier Dataset: Methodology and Baseline Performance

WTW targeted senior decision-makers rather than technology practitioners, pulling responses from executives in analytics, actuarial, and strategy roles. The data therefore reflects institutional intent and resource allocation rather than tool inventory, which is the right measurement when the question is whether analytics investment produces underwriting performance. Carriers can report high tool counts and low returns; the WTW methodology correlates investment orientation with actual combined ratio outcomes over a three-year window.

The six-point combined ratio gap is the headline performance differential, but the survey's value is in the disaggregation beneath it. Carriers at the analytics frontier show near-universal adoption of underwriting and pricing analytics: close to 80 percent rely on advanced rating and pricing models, with an additional 11 percent planning to implement them in the near term. The performance distinction between leaders and laggards is not primarily in underwriting analytics, where adoption is broad and relatively mature. It is in claims functions, where the data shows most carriers have barely started.

That segmentation explains why the combined ratio gap is six points rather than two or three. Underwriting analytics alone can tighten risk selection and improve pricing accuracy, but the largest loss cost and expense savings in a P&C book accumulate on the claims side: fraud detection that catches loss before payment, severity tools that reduce reserve volatility, straight-through processing that compresses cycle time and loss adjustment expense. Carriers that have deployed analytics across both underwriting and claims are the ones running six points better. Most carriers have only done the first part.

Claims Analytics Adoption: The Fraud and Severity Gap

The WTW survey is specific about where the claims analytics gap is. Fraud detection is the application most actuaries and claims leaders identify as the clearest analytics ROI case, and it is also the application where the gap between current deployment and projected adoption is most pronounced. Thirty-three percent of surveyed carriers currently use advanced analytics for claims fraud detection. The survey projects that figure reaching 65 to 70 percent within two years. That is a doubling of deployment in an application where most carriers acknowledge the ROI case is established.

Claims severity assessment is even less developed. Twenty-nine percent of carriers currently apply analytics to severity estimation. The same two-year projection puts that figure in the 65 to 70 percent range. Together, those two statistics define what most of the P&C industry has not yet done: apply statistical rigor to the claims functions that most directly drive loss ratios. A carrier writing $500 million in commercial auto premium and running a 72 loss ratio without severity analytics, shifting to a model that identifies high-severity indicators at FNOL and adjusts reserve estimates accordingly, does not close the gap in one year. But it does change the trajectory of reserve development in a measurable way within two to three accident years.

Straight-through processing sits at 14 percent current deployment, with 36 percent of carriers planning implementation within two years. Claim triage analytics is at 25 percent; subrogation identification at 20 percent; litigation probability modeling at under 15 percent; case reserving analytics at 20 percent. These figures describe an industry where analytics investment has concentrated heavily in pricing functions and barely penetrated the claims function, even though claims is where the loss dollar actually lands. The carriers running six points lower on combined ratio have penetrated both.

Human underwriting augmentation follows a different trajectory. Sixteen percent of carriers currently use analytics to inform human underwriting decisions at the point of risk selection. Sixty percent plan to prioritize that capability by 2028. GenAI adoption shows the fastest stated adoption trajectory: more than 50 percent of surveyed carriers are already implementing large language models in some capacity, with an additional 29 percent planning adoption within two years. But current GenAI deployment is largely concentrated in document processing and customer communication workflows, not in the core pricing and reserving decision systems that move combined ratios.

The technology architecture the survey surfaces is relevant for actuaries evaluating model governance requirements. Gradient boosting machines remain the dominant tool for underwriting analytics and fraud detection across the 59-carrier sample. That means most of the analytics ROI the leading carriers are capturing comes from well-validated, interpretable model classes that are compatible with ASOP No. 56 documentation requirements and state regulatory rate filing review. The carriers achieving the six-point CR advantage are not doing it with opaque architectures; they are doing it with gradient boosting applied to clean, structured claims and underwriting data, with feature importance outputs that actuarial staff can review and validate.

Claims Analytics Application	Current Deployment	Projected 2-Year Adoption
Fraud detection	33%	65-70%
Severity assessment	29%	65-70%
Claim triage	25%	Not specified
Case reserving	20%	Not specified
Subrogation identification	20%	Not specified
Straight-through processing	14%	50% (14% current + 36% planning)
Litigation probability modeling	<15%	Not specified

Barriers: Data Quality, IT Bottlenecks, and the Strategy Deficit

The WTW data on barriers to adoption is where the survey becomes most actionable for actuaries sitting inside carrier finance and risk functions. Forty-two percent of carriers cite data quality issues and limited data accessibility as significant obstacles. Inadequate IT support is cited as a major constraint by a similar share. These are not answers that point to a shortage of analytics tools or vendor options. The tool landscape for insurance analytics is well-developed. The shortage is in the data infrastructure that makes those tools functional at scale.

The strategy and training numbers are more revealing than the technical barriers. Only 20 percent of surveyed carriers report having a well-defined analytics strategy. Only 12 percent regularly provide analytics training to employees. A carrier that acquires analytics software without a strategy for deploying it against specific operational problems, and without training the underwriters and claims staff who interact with model outputs, will generate technology spend without generating combined ratio improvement. The 12 percent training figure in particular explains much of why claims analytics deployment numbers are as low as they are: adjusters who do not understand how severity models work, or why a triage tool is recommending a specific settlement path, will override or ignore outputs, negating the ROI and producing the kind of inconsistent performance data that makes model validation difficult.

A well-defined analytics strategy, which only one in five carriers has formalized, answers three questions that most carrier leadership teams have not resolved: which specific operating decisions should be model-driven versus judgment-driven and at what confidence threshold; what data inputs those models require and whether the carrier's data infrastructure can reliably supply them on a production basis; and what the governance and validation process looks like for model outputs before they affect underwriting or claims decisions. Carriers that have answered these three questions before deploying tools are the ones appearing in the top quartile of the WTW performance data. Carriers that deployed tools first and are working backward to answer the strategy questions are generating the 42 percent data quality complaint rate.

The IT bottleneck translates into something specific for actuaries. When carriers cite inadequate IT support as a barrier to analytics adoption, what they are typically describing is the absence of governed data pipelines: structured, documented, regularly audited flows of claims, underwriting, and external enrichment data that feed model training and inference reliably. An analytics model running on data pulled from a legacy system through an undocumented extract process is not a production deployment; it is a proof-of-concept that will fail the first time the extract format changes or the source system is updated. The carriers at the leading edge of the WTW performance distribution have built production-grade data infrastructure. The carriers citing IT as a barrier have not, and they will continue to underperform on both analytics adoption metrics and combined ratio until that infrastructure exists.

Top-Quartile Carrier Practices: Strategy, Training, and Data Governance

From tracking analytics maturity signals across twelve quarters of carrier earnings commentary and comparing them against the WTW benchmark data, the distinguishing practices of top-quartile analytics adopters follow a consistent sequence that the survey quantifies but does not fully explain. Strategy clarity precedes tooling. Carriers that begin by defining which underwriting and claims decisions should be informed by model outputs, and how those outputs will integrate with existing workflows, deploy tools against specific problems with measurable success criteria. Carriers that begin by selecting vendors and deploying models then figure out what problems to apply them to are the ones generating technology spend without performance results, and the ones accounting for most of the 80 percent below the six-point CR threshold.

Regular employee training is the second distinguishing practice, and the 12 percent training figure from the WTW survey represents one of the most consequential resource allocation gaps in the industry right now. An analytics model deployed into a claims or underwriting workflow without training the humans who interact with it produces adoption rates that undermine the model's performance metrics. Adjusters who distrust severity models override them on high-frequency decisions where the model has the most statistical power. Underwriters who do not understand how a risk-scoring tool is flagging adverse characteristics dismiss its recommendations on the accounts where it adds the most marginal value. Top-quartile carriers have solved this by investing in training before deployment and building feedback mechanisms that allow model performance data to flow back to the teams using the tools. The model improves because the human users understand what it is doing and report when it is wrong.

Governed data pipelines that actuaries can validate are the third element. This matters specifically for reserving and pricing actuaries whose work product is subject to regulatory review and ASOP compliance requirements. A carrier whose pricing actuaries cannot access documentation on the data inputs, feature engineering decisions, and validation results of the underwriting analytics tools affecting their book of business is operating with a blind spot that ASOP No. 56 specifically requires be addressed. The carriers running six points better than their peers have solved this integration problem. Their actuarial functions have visibility into what the analytics tools are selecting and rejecting, and that visibility feeds into pricing assumptions and reserving judgments.

GenAI's role in the top-quartile carrier profile is supporting rather than primary. The majority of carriers already implementing LLMs are applying them to document processing, policy language analysis, and customer communication workflows, not to the core risk selection and severity estimation decisions that drive combined ratio outcomes. The leading carriers understand that GenAI accelerates workflows around the core analytics decisions; it does not substitute for gradient boosting models trained on structured loss data for the decisions that move loss ratios. The 60 percent planning to prioritize AI-augmented underwriting by 2028 will discover this sequencing requirement when they get there: the underwriting augmentation ROI depends on having clean structured underwriting data, which requires the same data pipeline investment that the 42 percent data quality complainants have not yet made.

The 2030 Projection: Morgan Stanley's Two-Hundred-Basis-Point Expense Ratio Estimate

Morgan Stanley's analysis of AI's long-term impact on commercial, specialty, and reinsurance carriers projects a two-hundred-basis-point expense ratio differential by 2030 between carriers that deploy analytics and AI at scale and those that do not. The non-AI carrier trajectory produces a 30.5 expense ratio; the scaled-adopter trajectory yields a 28.5 expense ratio. The operating margin differential is approximately 180 basis points: 15.6 percent without AI versus 17.4 percent with it.

Applied to the WTW data, that projection describes how the current six-point combined ratio gap compounds over time. The WTW data measures performance during 2022 to 2024, when the leading carriers had already invested and were capturing early returns on those investments. By 2030, the analytics leaders will have compounded four to six additional years of model improvement, data accumulation, and workflow optimization on top of the base they built in the early 2020s. The 200 basis point Morgan Stanley expense ratio differential captures only the expense side of that compounding; it does not include the loss ratio improvement that the WTW data attributes to analytics leadership, which is where the larger share of the six-point CR gap originates.

The compounding dynamic has a specific interpretation for actuaries. A carrier that achieves a six-point structural combined ratio improvement through analytics investment does not simply run at a better result for a single year. It enters subsequent underwriting years with a better risk portfolio, selected more precisely at point of underwriting; better reserve development patterns, from claims severity and STP analytics; and lower expense ratios, from process automation and reduced LAE. All of these reinforce the next year's starting position. Carriers that invested in analytics between 2020 and 2024 will enter 2025 through 2030 with structural advantages that are increasingly difficult for later adopters to replicate quickly, because the competitive moat of a functioning analytics program is partly in the data accumulation and model iteration history that took years to build. A carrier starting in 2026 is not starting from the same position as a carrier that started in 2021; it is starting from a position four to five years behind on training data, model validation cycles, and organizational muscle.

For CFOs and actuaries defending analytics investment budgets in 2026, the WTW and Morgan Stanley data together supply the benchmark arithmetic that internal business cases have been missing. Travelers' $1.5 billion annual technology commitment, with AI spend more than doubling over eight years and 20,000 employees using AI tools, shows what scaled commitment looks like from a carrier already in the analytics leadership cohort. The WTW six-point CR gap is what that commitment has been producing at the cohort level, not just at the Progressive or Travelers outlier level.

Actuarial Implications: Reserve Development Divergence and the Benchmark Problem

The six-point combined ratio gap between analytics leaders and laggards creates a specific and underappreciated problem for actuaries at both types of carriers: the industry benchmarks that anchor pricing assumptions, reserve selections, and competitive rate adequacy assessments are increasingly unreliable as the performance gap widens.

Pricing actuaries at laggard carriers who calibrate target loss ratios against industry combined ratio benchmarks are calibrating against a number that averages across carriers with materially different risk selection quality. A carrier running a 65 loss ratio through analytics-driven underwriting writes a meaningfully different distribution of risks than a carrier running a 72 loss ratio through traditional methods, even within the same line of business and the same state market. Using industry benchmark loss ratios to set rate adequacy for the laggard carrier overstates adequacy because the laggard's book includes risks that the analytics leader has already rejected or priced more precisely. This is a rate adequacy trap that compounds over renewal cycles: the laggard carrier prices to industry benchmarks, attracts the risks the analytics leader declined, and ends up with a book that is worse than the benchmark, not average to it.

Reserving actuaries face a parallel challenge from claims analytics deployment. A carrier that adds fraud detection and severity assessment analytics, even at the 33 percent and 29 percent adoption rates the WTW survey reports for laggard-to-average carriers, will begin generating claims development triangles that diverge from its own historical experience. Fraud detection changes the composition of claims that reach payment; severity analytics changes the distribution of settlement amounts and timing on claims that do. Loss development factors calibrated to pre-analytics experience will be wrong, in ways that depend on which applications were deployed, when, and on what book of business. Actuaries at carriers adding these tools need to segment development triangles by analytics maturity period and build explicit adjustment factors for the behavioral change in the underlying claims process. Standard triangle methods will not detect the shift until several accident years have developed enough to reveal the pattern, by which point the reserve error is already embedded.

ASOP No. 56 compliance is also directly implicated by the WTW findings on strategy and governance. The standard requires documentation of model assumptions, limitations, and validation procedures for models that inform actuarial work products. For analytics leaders with governed data pipelines and regular model validation cycles, this documentation exists and is updated. For carriers where analytics tools were deployed by vendor implementation teams without involving the actuarial function in validation design, the ASOP No. 56 documentation may be absent or incomplete, even as the models influence underwriting decisions that flow through to rate filings. State regulators in Colorado, California, and New York are increasingly asking specifically about model documentation for algorithmic underwriting decisions. Carriers that cannot produce this documentation face filing delays, objections, or conditions that slow their deployment timelines, which is itself a competitive disadvantage relative to carriers that built governance infrastructure before deployment.

The 29 percent claims severity assessment adoption rate carries a specific reserve adequacy implication. Carriers that do not apply analytics to severity estimation are selecting development factors based on historical patterns that may not reflect the current claim environment: elevated litigation costs, nuclear verdict exposure, and supply chain-driven repair cost inflation each affect severity in ways that gradient boosting models trained on recent data can detect and incorporate faster than traditional actuarial judgment applied to triangle patterns. Carriers that have deployed severity analytics are identifying adverse development signals earlier, adjusting reserves before the development fully emerges in the triangle. Those that have not are at risk of the reserve development surprise that always follows extended periods of severity inflation that ran faster than IBNR selections assumed.

Competitive benchmarking is the third area where the WTW data changes actuarial practice requirements. Patent filing activity, which has tracked closely with analytics investment depth at leading carriers, now shows GenAI patents at 31 percent of insurer filings, concentrated among a small group of large carriers. The patent signal confirms what the WTW performance data shows: analytics investment is not distributed evenly across the market, and the carriers at the front of the distribution are pulling ahead faster than the aggregate adoption statistics suggest. Actuaries performing competitive rate adequacy analyses should disaggregate benchmark data by analytics maturity where carrier-level detail is available, particularly in commercial and specialty lines where the WTW survey found the smallest but fastest-growing analytics investment programs.

The WTW survey's value is not in the headline numbers, which are significant, but in the anatomical detail beneath them: the 33 percent fraud detection rate pointing to where most of the untapped ROI sits, the 12 percent training rate pointing to the root cause of slow adoption, and the 20 percent strategy rate explaining why tool investment has not translated into the six-point CR advantage that the leading cohort is demonstrating at scale. Actuaries setting pricing assumptions, selecting reserves, or defending analytics investment budgets have a primary-source benchmark from 59 carriers with a three-year performance record. The arithmetic of a six-point structural combined ratio advantage compounding into a 2030 expense ratio differential of 200 basis points is not an abstract projection. It is where the industry is visibly heading, for the carriers that have invested, and not heading, for those that have not.