Buzzwords can gain momentum quickly in IT, and the latest technology buzz is around "Big Data." Netflix's Jeremy Edberg, in an article for InformationWeek (a UBM TechWeb brand), recently defined big data as "the tools and processes of managing and utilizing large datasets." But for insurers looking to build smarter claims organizations, the tools are much more important than the size of the datasets themselves, experts say.
"There really is a ton of data in insurance," points out Aaron Fidler, VP at Applied Predictive Technologies, a Washington, D.C.-based provider of analytics software and consulting services. But, "There's always that question: Is more information going to be more valuable? Given what we've seen, it's critical to understand what is going to be helpful and what is noise."
To gain that understanding, most U.S. insurers are building proprietary data collection, storage and analytics capabilities, adds Craig Beattie, a London-based analyst with Celent. "Certainly big data has a role to play here," he says. "In the U.S., there's a little bit more proprietary analysis because insurance companies have invested in the infrastructure to do that."
Mike Cesinger, VP of claims for Plymouth Rock Assurance ($1.2 billion in claims expense reserves, 2010), speaks with pride about his company's in-house analytics unit. The Boston-based carrier, which offers personal lines auto and home insurance, evaluates claims not by bolting on new data sources to its existing ones, he explains, but by proactively looking for the information that most accurately represents the profile of a claim.
"There's a shifting set of data elements that we look at," Cesinger says. "We're proud of the fact that we looked at not just the nature of the claim, but the nature of the risk — not just the insured's claims history, but garaging locations and undeclared operators, among other things. We're taking a more holistic approach to evaluating individual claims and not just triaging them based on the description of loss."
An Accurate Bucket List
At Plymouth Rock, claims are processed into one of three buckets, Cesinger adds: One bucket contains claims headed for the special investigations unit (SIU); another contains those that are potentially problematic, but don't necessarily need the SIU's attention; and the third is for claims that are unlikely to cause any problems. As data and analytics tools have proliferated outside of the IT department, he says, it has become easier and faster for the insurer to get a claim into a specific bucket — or move it from one to the other.
"The ability of our analysts to massage our data and let it take us where it needs to in a short period of time — but still be nimble in what our response might be — has been a huge boon," Cesinger says. "In the past, when you had your IT department write a query, you only got an answer to that query a couple weeks later. Now you've got the ability to go in and identify visual patterns and use neural networks to identify associations of data immediately."
Like many companies, Plymouth Rock has been applying its new analytics capabilities to social media data, Cesinger reports. Policyholders can expose a surprising amount of information via social networks, he notes. "It's interesting when you have someone who reports that their car was hit while unattended in the supermarket, but at the same time that was supposedly happening, they've got a Facebook post about how drunk they are," Cesinger says. "Young people have much more desire to be totally forthcoming in their social media posts, and that sometimes can be gold for an insurance investigation."
But that doesn't mean that the only use for analytics is to police policyholders, Applied Predictive Technologies' Fidler stresses. "Right now there's a focus on how we reduce fraud, but also on retention and interaction, as we think of ways to interact with our customers at the point of a claim," he says.
[For more about the 4 Ways Big Data is Changing Claims, see Nathan Golia's related story.]