Nowhere is the insurance industry’s notorious resistance to change more evident than in how they fiercely it clings to antiquated IT architectures. Carriers adopted mainframes in the ‘80s, built policy and claims engines on top, then lagged years behind just about every other commercial enterprise in embracing client-server computing, then cloud computing. You can hardly blame them. The systems that manage policies, claims, and billing are the heart and lungs of an insurance operation. They are how an insurer checks eligibility against underwriting rules and prices risk; how it sets reserves, prioritizes claimants, and assigns claims to adjusters; how it tracks when premiums are due and how long they’ve been overdue.
Those systems, in turn, are integrated with hundreds of other applications. Ripping out and replacing them with modern versions to wring out incremental efficiencies not only risks massive business disruption but loads the income statement with enormous implementation costs for years. Insurers, wired by design to fret over downside than to chase upside, stick to proven systems of record….until the risk/reward calculus definitively flips and running mainframes poses more downside than installing a modern core because there is no one left to maintain layers of custom hardcoded business logic as all the COBOL programmers have retired; because rigid schemas and nightly batches coarse-grain pricing and risk scoring to such an extent that loss and expense ratios are at risk of bloating; because the process of quoting policies and cycling through claims is so slow that customer acquisition stalls and retention suffers.
The industry had just about reached this point when Guidewire released ClaimCenter (claims intake, adjudication, processing), PolicyCenter (quoting, rating, underwriting), and BillingCenter (invoicing, payments) in the early/mid-2000s. These three Centers and the InsuranceSuite they together comprised, offered a modern, more flexible alternative to decades-old mainframes. But, because they were hosted on-premise, insurers, preoccupied though they were with the day-to-day business of pricing risk, were also forced to assume the burden of an IT administrator.
Across a growing number of industries, IT functions were increasingly offloaded to cloud infrastructure providers during the 2010s. True to form, the Tier 1/2 P&C carriers1 that Guidewire served were at first hostile to the whole idea. Guidewire’s first cloud platform, launched in 2016, lacked the multi-tenant design and service-oriented architecture we now take for granted in modern cloud platforms. Carriers merely lifted on-premise monoliths from their servers to Guidewire’s, with custom configs in place and everything. It wasn’t until 2019, with former Salesforce executive Mike Rosenbaum at the helm, that Guidewire ripped off the band-aid and built a true cloud foundation on top of AWS.
Guidewire broke the InsuranceSuite monolith into microservices; introduced more shared services, like logging and tracing tools that were independent of the rules and ratings engine that were separate from the UX modules, across customers; embedded analytics and data curation capabilities; and only permitted custom configurations according to certain standards.
At its core, GWCP is built on off-the-shelf services and the data platform on top relies on lots of open source data pipeline tools. Carriers could technically build their own mini-clouds on AWS, but it just wouldn’t be worth the effort for any single insurer. Better that Guidewire scale investments in P&C specific objects and data curation across its 400 carrier customers and share the resulting scale economies.
In shifting to the Guidewire Cloud Platform, or GWCP (”GCP” was already taken!), carriers surrendered some of the customization they enjoyed with self-managed systems. But they gained far more in return. Upgrades that used to be available only every two to three years (and which carriers often delayed due to cost and time constraints), were rolled out every six months. Bespoke point integrations were supplanted by consistent APIs that allowed third party “insurtechs” to plug into the Guidewire mothership with less custom code. Customer and agent portals that were previously built from scratch could be assembled with ready-made building blocks. Overnight batching was replaced by near real time streaming of data that could be fed into iteratively tweaked machine learning models to inform underwriting decisions and automate claims processing. As management explained in last year’s Investor Day:
we uniquely solve this problem across the entirety of an insurance life cycle, okay? We have an application that covers everything from the definition of a product, to the distribution of that product, to the quoting and underwriting of that product, to managing the policy once it’s in force and then managing a claim if and when it occurs. And we’re collecting all of that information that we can glean from that business process, putting it into our data platform and providing it to the insurance companies so that they can make this loop operate more and more effectively and more and more efficiently over time.
Beyond the capabilities it unlocked, GWCP offloaded messy IT work from carriers who had neither the talent nor cultural constitution to maintain a modern core to a dedicated vendor who spent the last twenty years doing nothing but that. In return for taking on this work, Guidewire charges subscription fees that are 2x to 3x greater than what a carrier pays for on-premise term licenses and maintenance. But that 2x to 3x uplift reflects “fully ramped” revenue baked into signed deals, which builds gradually over a period of up to five years through phased rollouts across InsuranceSuite modules (ClaimsCenter, PolicyCenter, BillingCenter), business lines (commercial, personal, auto, etc.), geographies, and add-on products like predictive analytics and digital interfaces.
In the meantime, with the cloud transition still in its infancy and the industry watching early adopters with hawklike intensity, it was critical that Guidewire not fuck up, that literally every cloud implementation succeed without exception. And so, even as it exchanged pure margin term licenses for subscriptions fees layered atop cloud infrastructure that was nowhere close to being fully utilized, Guidewire also deliberately overstaffed its services division.
The result?