As more non-traditional players enter the insurance market andtrends like InsurTech advance, one thing is certain: Insurers mustinnovate to remain competitive.

But there are many new ideas to choose from, and deploying a newinitiative can be quite risky, both financially and in terms ofbrand reputation — especially if it doesn't perform as expected inmarket.

Before investing in broader rollout of new programs,organizations first need to determine which will be the mosteffective and what factors drive their success.

Many organizations across industries leverage experimentation tocritically evaluate new initiatives and pinpoint their incrementalimpact on the business overall. Although companies worldwide areincreasingly adopting a testing discipline, many insurers have notyet fully capitalized on this approach to linking cause and effect.One reason for this inertia is that historically, some insurershave not perceived certain business actions as testable, due tofactors like complex, state-specific regulations.

One proven approach for insurers to unlock insights inhistorically under-tested areas is to first identify instances ofnatural variation — cases where testing was not conductedintentionally, but a measurable change occurred in the business.Within the day-to-day operations of an insurance company, there aremany footholds of this kind that can serve as “gateways” toexperimentation, the first step in enhancing insuranceorganizations' analytic capabilities.

Continue on for examples of key gateways that insurers can useas starting points to jump-start their analytics programs.

No. 1: State Regulations

State regulations enforced by each state's Department ofInsurance mean that a single insurer often needs to run itsindividual states as somewhat distinct businesses. While this posescomplexities, it also means that changes like product linemodifications, rate adjustments and altered risk scenarios areimplemented differently in a subset of states, providing potentialgateways to experimentation.

While industry regulations across state lines may at first seemlike an obstacle to testing, the natural variance in differentinitiatives and programs by the state can create opportunities fordata-driven program management. For example, if an insurerintroduces a new type of policy in some states, it can compareperformance across key metrics to similar states without theintroduction to isolate the new product's true impact.

No. 2: Innovation

Insurers are increasingly dedicating significant resources toinnovation. Drawing on ideas from the field, policyholders andcompetitors, teams are continuously working to compile the best newinitiatives for their organization, pilot them and refine theirgo-forward strategy. These new programs are natural gateways toin-market testing.

Telematics is one example of this kind of innovation. Telematicstechnology allows insurance organizations to develop a clearer viewof policyholder behavior, then incorporate this knowledge into riskassessments and pricing. Beyond using these insights to betteralign premiums with risk to enhance rate fairness, insurers canalso assess other key program considerations by measuring theimpact of introducing a telematics platform. These considerationsinclude how the technology should be introduced and communicated toadditional states, markets and policyholders; which promotions,such as “Safe Driver” discounts, are most impactful in drivingusage and encouraging safer behaviors; and how this offeringultimately affects policyholder retention and satisfaction. Fromthere, insurers can tweak their program implementation to reach thepolicyholders with whom it will be most effective.

Because innovative programs like telematics are often introducedto markets in phases, insurers have the opportunity to save bothtime and resources by fine-tuning these new programs every step ofthe way before investing in broader rollout.

No. 3: Opt-in Programs

Another gateway to experimentation is through programs that areover-subscribed — specifically, initiatives where more people wantto play a role than there is capital to fund them. Because of theamount of interest in these programs, there may be self-selectionbias; that is, the characteristics of agents or policyholders thatopt in may impact whether the initiative itself achieves theintended affect.

One example of such an opt-in program could be a new,sales-focused agent training program. Consider a scenario in whichan insurer developed such a program, and wanted to determinewhether it drove a large enough increase in premiums to justify thecost associated with offering the training program more broadly.Although premiums rose after the program launch, it might bedifficult to determine if this positive performance was driven bythe training program itself or other factors, such as whether theagents that opted into the program were already high performers.One way to isolate an individual program's incremental impact is toanalyze the performance of agents who opted into the programagainst the performance of a group of agents of similarcharacteristics who did not participate.

Implementing strategies that might shift behavior — such astrainings and new incentive structures — is not without risk. Byidentifying “natural experiments,” analyzing existing variation asthe first step in developing a broader analytics program, insurerscan use measurement of past programs to inform their future rolloutdecisions.

No. 4: Variation in Sales Force Deployment and CallCenter Activity

Natural variation resulting from changes in sales forcedeployment and call center activity can also serve as a gateway toexperimentation.

Consider, for example, an insurer that routed policyholders witha high likelihood to discontinue their coverage to a new specialtycall center, in hopes of mitigating attrition. The specialty callcenter was not large enough to handle all such calls, resulting intwo similar groups of “high attrition risk” policyholders: Thosethat were routed to the specialty call center, and those that wereserved by standard call center representatives. Measuring theretention rates between the policyholders who were routed to thespecialty call center and a carefully matched subset of those whowere not, the insurer could quantify the specialty call center'sattributable impact on retention.

Similarly, an insurer leveraged past natural variation in theirsales team's outreach to agents to determine the incremental impactof this engagement on quote and application rates. This measurementalso identified which types of interactions generated the greatestincrease in production and which types of agents benefited mostfrom this outreach. Armed with these insights, the insurer couldimprove its deployment of sales team effort, optimizing resourcesby prioritizing the most impactful outreach.

These gateways are a critical starting point for insurers toadopt a culture of testing and spur innovation. By capitalizing onthese opportunities to unlock valuable, data-driven insights,insurers will be empowered to introduce initiatives on an optimalscale and refine new business strategies for maximumprofitability.