23. AAAP Case Study in Financial Services
Best offer for hot prospects
Business Issue
To identify potential clients among all the prospects
interacting through WebSite channel to ask for a quotation.
Identify the best offer to address:
• «hot» prospects to turn quotations into policies
Work on the last batch of quotations (typically week),
focuses on those still alive, and then score the success
probability.
The high potential, are assigned to channels for further
contact.
Accenture Approach
1.3M Quotations as input to the analysis.
Main challenge noisy data.
Eg: husband ask for a quote with his name a fake plate
number, compares a few offers, then get back for the final
quotation using wife name real car data.
Lot of data cleansing, complex data preparation, to define
the Potential Sale Record.
Input data: Customer, Relationship with firm, Quotes, Car
Logistic Regressions for probability scoring, Modeler engine
Key Drivers:
• F customer
• Price<400€
• Cylinder Capacity < 1200cc
• Closeness Policy expiration date (<7d)
• Previous quotations >2
• Other Policies >1
Gains: the 10% top ranked quotations contains 50% of the
conversions
Next step: Cluster analysis (k-means) on existing clients,
using internal and external insurance information and
banking behaviour information to segment and profile
clients.
Client
Italian Insurance company, part of one of the
largest Banking groups in Europe