Constrain your analytics initiatives
We’re going to get ourselves in trouble with this one. We have an ugly observation to make about insurance IT. Here goes:
Your company spends a ton of money on analytics. But your company will be too market-constrained to respond to 80% of the insights that your advanced analytics reveal.
We’ve seen this over and over. An analyst poring through the data finds some sub-coverage that is generating outsize losses, or some market segment that is tragically unprofitable, or some growth opportunity driven by macroeconomic developments… only to find out that the front line guys had already spotted it, vetted it, and found it to be non-actionable.
When this happens, it seems like a depressing realization. “We just can’t do anything right,” your people say. But here’s the way to look at it: this is a function of a super-mature, fully saturated market, where every degree of freedom is already constrained by competitive forces. It’s not a problem with your company. The problem is that you’re in a damn tough industry.
It’s like living in a major city and thinking an old-west style land grab is possible. That’s foolish. The sophisticated city dweller understands his constraints, and works artfully to thrive within them.
It is times like these when you need to remember why you exist in the first place: to provide superior service to the clients who you have relationships with. That is why people pay you — it’s where the river of cash flowing through your company has its origin. Protect that river. Don’t let it be dammed up or diverted. And remember that your market-facing people are the ones who know best where opportunities lie.
But there’s more. Consider the cost imposed on your organization by the failed 80% of insights. The costs are not only the explicit IT budget costs you see on your capitalized asset ledger, but also the disruption caused to the normal flow of your producers’ work. By forcing them to collect all that data, you have impeded the river of cash.
How, then, should you pursue analytics initiatives? We suggest using a scientific method: form hypotheses, then test them. Instead of easter-egging about (searching for insights in a haphazard way), let the people closest to the action form their top hypotheses about which business to grow and which to cut. Then come up with targeted experiments that will confirm or deny those hypotheses. Yes, these experiments are likely to be data-intensive, but not nearly so intrusive as the NSA-style “collect it all” approach that you were using before. You may find that a much narrower subset of data will do fine. And you may also find your front line people eager to provide that data.
Embrace your constraints. They often allow you to restate problems that require complex technical solutions into problems with simpler solutions.
