[GWRE – Guidewire Software] Move slow and maintain things

If I had to force rank industries most resistant to new technology, P&C insurance would be right up there with other notoriously innovation averse sectors like local government, life insurance, education, and healthcare.  To some extent, this makes sense.  Given the critical nature of their services, insurance carriers should be culturally biased towards minimizing downside risk.  There are serious consequences to moving fast and breaking things.  But after trundling along for decades with no imperative to adapt, insurers have started to realize that their antiquated systems are a competitive liability in a world where policyholders – who perceive insurance companies to be culturally stultified institutions, as monolithic and constipated as the technology they run on – expect responsive, consumer-grade experiences. 

The last 6-7 years has witnessed an explosion of InsurTech startups, some of whom are competing against incumbents and others who are selling applications to them.  Something like $4bn of funding has made its way to companies addressing insurance over the last few years.  “InsurTech” has become its own species under the fintech genus.

Here is a sample of the landscape, pulled from Guidewire’s 2018 Investor Day Presentation:

Within the Digital Attackers pool are companies finding new ways to change incentives and pool risk.  They are leveraging machine learning to reduce costly manual processes and using novel sources of data to more efficiently underwrite risk.  Most of them interact directly with consumers through sleek user interfaces and position themselves as the empathic, responsive counterpart or alternative to the faceless incumbent monoliths that policyholders have long despised.  

Lemonade promotes better claims behavior by donating whatever money is leftover after the company has taken its fixed percent and claims have been paid, to a charity of the policyholder’s choosing.  By aligning policyholder incentives with its own, Lemonade reduces fraud propensity and can pay off a quarter of all claims instantaneously (apparently, policyholders hate their insurers so much that a significant minority see no problem inflating their claims).  The company uses bots to file paperwork with regulators and to cancel a prospective policyholder’s existing policy with an incumbent carrier, obviating the teams of people that most insurance companies hire to handle this kind of grunt work.  Combining behavioral economics and machine learning to align incentives and reduce overhead, Lemonade can offer insurance at prices so low – starting at $5/month for renters’ insurance, $25/month for home – that it (supposedly) becomes an impulse buy. 

Rather than target large categories like auto insurance whose keywords are claimed by incumbents with scale, Next Insurance aggregates niche commercial risk pools like contractors, therapists, and yoga instructors that are overlooked by traditional insurers.  Rather than having prospective insureds answer 60 arcane questions about their home or condo, Hippo has them answer just three and gets everything else it needs from dozens of public data sources (property records, aerial photography).  Rather than bluntly committing policyholders to month-long payment terms, companies like Zego, Cuvva, and Slice offer pay-as-you-go on-demand insurance by the hour or day for rideshare and delivery drivers and homesharers.  All these companies offer sleek user interfaces and a paperless flow to onboard and support policyholders.  They all use data and machine learning to limit manual processes and control risk.  Next Insurance compresses a weeks-long application crucible down to 10 minutes.  93% of its customers never talk to a human.  Hippo relies on smart home sensors and IoT devices to preemptively detect damage and uses aerial images to monitor homes.

Most of these start-ups either tailor general tools for the insurance sector or streamline customer acquisition.  Very few are selling back-end systems.  Panelists at various InsurTech conferences will pontificate on disruptive business models and innovative ways of applying horizontal technology, but virtually no one talks about replacing core systems.  And no wonder – core systems are hard!  Really hard.  It took Guidewire nearly 3 years to build its first minimally viable product, ClaimCenter, which it launched in 2003 and leveraged to cross sell additional applications addressing policy management (PolicyCenter, 2004) and billing (BillingCenter, 2006)¹.  These three solutions, combined with a rating engine that rapidly updates pricing in response to new information, make up Guidewire’s InsuranceSuite.  You can think of Guidewire’s InsuranceSuite as modern ERP for the insurance sector, a mission critical transactional system of record.  It is the leading alternative to creaky 30-year old systems, programmed in COBOL and running on mainframes, that most large carriers still use to quote, issue, and maintain policies; bill their customers; and process and pay out claims. 

Neither InsuranceSuite nor its component pieces can be sampled the way viral tools like PagerDuty or Jira can today.  They require the whole enterprise customer to commit to a massive, time consuming transformation project.