Phil, Director Customer Insights at a $15bn Regional Bank.
Phil heads a team of Data Scientists who have a great track record building predictive models that score customers’ propensity to buy – or abandon – at both the product and relationship levels. They know model accuracy directly impacts financial results, and are seeking new ways to gain insights ahead of competitors.
Phil’s team recognizes their models are only as good as the target definition and the predictive variables they have on file. IT wants to invest in unstructured social data, but Phil believes they can get more value from the data they already have.
|Problem:||Customer Sales and Attrition are not separable from product substitution in Phil’s data, clouding the target models predict. |
Acquiring new data sources is costly with uncertain results. Gaining more insight from existing data is what Phil really needs.
|Solution:||Phil implemented FlowTracker Cloud SaaS, backfilling 36 months of time series analysis. Existing models were updated to redefine sales and attrition targets as pure “new” and “lost” business. Phil’s team also added the new customer behavior events into their models as predictors.|
|Outcome:||Within 120 days a full 36 month history was available for inclusion in predictive modeling. Phil’s team discovered their new models gained 30% in accuracy due to target redefinition and FlowTracker customer behavior events had significant predictive power when added to existing models.|