Usage of Advanced Analytics is fast becoming more pervasive amongst Insurers in India. Players with bigger market share and those that have been around in the industry for quite some time have been really aggressive in this play with in house analytics teams working on different business initiatives. At the same time we have insurers that have entered the market in last 3-5 years and are still evolving towards setting up mature distribution,sales process and finding right product mix. For these players all this hype around Predictive Analytics may not seem to be a priority although they may lend keen ears to it.
Is it right time for them to adopt advanced analytics into business process? Some barriers one will encounter in this journey
- Consolidating data across silos
- Data quality
- Changes to business process and IT systems
- Training of business, operations and technical teams.
All these might pose substantial barriers to surpass and a first successful integration may take anyway more than 5 to 6 months. In light of this, one would rather focus on business priorities, strengthening sales and distribution channel and going after market share.
Is there a way we can adopt advanced analytics avoiding much of overheads?
It is also a fact that for most people advanced analytics is equated with Predictive Analytics where one is making predictions about customer churn, cross sell purchase etc and deliverable is an algorithm which outputs customer scorecard or propensity. Advanced analytics has more to it than just predictive analytics. What else can we do?
Customer Segmentation: This may seem plain simple Business Intelligence report to start with especially with defined set of segments in mind. But given that your business is collecting various different data points one may miss out on finer segments. At the same time you may not always find out interaction between different variables. There are unsupervised clustering techniques like K-Means, Fuzzy C and lot more that allow you to do segmentation more effectively without having a particular profile in mind. Once we get these segments and profile these out, business can devise right marketing or sales strategy for each segment.
Understanding Customer Profile for Events: Events like Churn, Cross Sell purchase are possible to predict with historical data. Without building a predictive model one can study segments that show higher churn or most cross sell. Decision tree techniques like CHAID, C&RT are effective techniques in this regard. Results from this can be interpreted by marketing team to target right profiles across different channels, Renewal team can devise right retention strategy for different segments, and Sales team can begin sourcing right set of profiles that are needed for healthy business.
Results of above studies can provide useful insights to business users. Think of it as strategic insights stopping short of technical implementation.
The post Beyond BI, Before Predictive Analytics appeared first on Analytics India Magazine.