Insurers for some time now have recognized the potential of big data to provide a strategic edge. But most find it challenging to capture this potential, in part because of the vast size and tremendous growth in both the volume and variety of existing data. Software intelligence is part of the key to capturing this potential, as I discuss in this Technology Vision for Insurance podcast “The Intelligent Insurer”.
How software intelligence works
The idea is that machines can use data to learn and make decisions. An intelligent thermostat, for example, would turn off the heat in an empty house, and turn it on again in time for the family’s return—not because it was programmed this way, but because it learned the behavior of the users, and created rules based on their habits. This example is straightforward and obvious, but new generations of intelligent software will “discover” complex, unexpected connections in existing data and learn from its experiences. For example, whenever the homeowner turns on the treadmill, she invariably turns down the thermostat and takes a shower afterwards. Intelligent software discovering that pattern can lower the thermostat and prepare for hot water demand, associated with treadmill use. Other associations, different for every individual, would also be discovered.
These examples illustrate some basic applications of software intelligence. Future developments in cognitive computing will expand a machine’s inputs, analysis, and ability to take action.
How can we use it for insurance?
Software intelligence can provide operational intelligence and innovation for insurers to have a real impact on the business.
Research suggests that by 2020, there will be more than 40 billion terabytes of data available. This is far too much for humans to identify, capture, categorize and analyze, and without support would not provide much value to insurers. But with improvements in storage capacity, processing power, and data science, intelligent software can handle data at this scale. It can also support or, in many cases, make decisions while learning to become more efficient and effective in its designated applications.
There are many potential applications for insurance, from automating rules-based decision-making to enabling highly customized and personalized customer interactions. Intelligent software can be leveraged to support sales processes, underwriting decisions, product configuration, claims segmentation and handling, and fraud detection among others.
Current market examples
Already, US firm Picwell is providing health insurance advice using intelligent software to rate health plan options by individual need. With over 900,000 variables affecting plan selection, the Picwell platform combines machine learning with predictive analysis and behavioral economics to make sense of the vast quantity of data involved.
Mark Halverson recently mentioned The Climate Corporation, which uses intelligent software to examine weather and related agricultural data, and provide insurance addressing farmers’ financial exposure to crop performance. This company handles more than 50 terabytes of live data through intelligent software, well enough that the US Department of Agriculture authorized it in recent years to administer federal crop insurance policies.
I think other insurers can follow the lead of these pioneering companies, implementing intelligent software to make big data not just a part of their operations but a leading source of innovation and strategic success. In my next blog, I’ll talk more broadly about how collaboration that brings together human skills and machine competencies will reshape the industry in coming months and years.
If you’d like more information, please visit Accenture’s Tech Vision for Insurance.