More insurance companies are working to achieve "testing nirvana"—a state of frequent experimentation—for accurate insight and causation into consumer behavior, says State Farm’s Director of Analytics Andy Pulkstenis.
Insurance carriers should always conduct experiments to quantify hypothesis before making permanent changes to website design, claims and other areas affecting business strategy, Pulkstenis told the crowd at last week's Insurance Analytics Synmposium, presented by INN
“People are weird and complex,” Pulkstenis says. “That’s why we test.”
On the road to testing nirvana, insurers typically go from no experimentation to A/B testing, Pulkstenis says. For example, running two versions of a web page or app feature to see which performs best. Many organizations, however, stay stuck in the A/B testing stage and fail to progress beyond it.
“Going from nothing to A/B testing is so powerful that insurers grow content with their level of experimentation,” he said. “Many carriers also work with vendors that only know A/B testing and advertise it as the best method there is.”
As it stands, State Farm has businesses all over the experimentation map, Pulkstenis notes. But a more mature culture develops customer segmentation practices, he says. Insurers group customers by age, gender and location to gauge trends. The result is a broadened understanding of customer behavior, which leads to less guess work and a higher success rate for implemented changes.
“The job is not done when experiments go live. You have to monitor tests, be ready to share results and educate your peers,” he said. “Start small. Predictive modeling is not new to the industry. Don’t ignore testing constantly because it seems too complicated.”
