Portfolio Analytics: What can the data tell us and how can we use it? Portfolio Analytics Panelists David Johnson - Managing Partner - Cane Bay Partners Chris Corcoran - VP Risk Management - MacFarlane Group Rich Alterman - SVP Business Development - GDS Link Greg Rable - Founder/CEO - FactorTrust Mark Doman - EVP Business Development - eBureau What does the data tell us about... • Successful analytic foundations • Key metrics and real life analytics application • Successful implementation processes What does it take to be good at analytics and portfolio management? Culture Investment Analytics Technology People Partners Data Databases Analytic tools Measurement systems Culture/People What kinds of people and skill sets are important? Visionary & progressive leadership Business, operations & technical competency Data analysis Database expertise Technology savvy people Internal or external statistical/statistician resources Legal & compliance What do we mean by analytics? Does that mean using rules or scores or both or what? Scores and rules are both important and come out of “analytics” Rules based on known historical data can be implemented with less expense Well developed scores are more statistically sound and can incorporate many variables Ease of implementation is an important consideration The solution(s) that drive the best results should drive the decision % of Funded Loans w/ this Message Code 28.5% What key metrics are companies focusing their analytic resources around? Lead flow Settlement rates Redirect rates Renewal rates eSignature rates Charge off rates Conversion rates Gross & net revenue Cost of acquisition Lifetime customer value FPD Servicing cost What type of decisions are lenders using analytics to make? SEO Renewal loan amount Marketing Customer care Fraud prevention Call routing Lead purchasing Collection Bid price Debt sale First loan amount What are some key components to a good analytics process? Really good planning Knowing what you want to achieve Measurement on how you are doing Listening to the data - it may not be what you think A-B testing “Vintage” reporting - changes and results can be connected Continuous improvement A commitment to lose some money for the sake of learning There is a large group of leads that score well but fail the current decision process. Look for way to relax or eliminate certain rules and profitably grow the portfolio # Leads Worst Score Best Score What types of data are lenders leveraging in their analytic processes? Historical results data ACH data Device ID data Banking data Geo location data Demographic data Velocity data Credit data Verification data Social media data Stability data What are the important factors to consider if you are planning on developing a score? A good definition of what the score should predict Development data that is accurate - garbage in = garbage out Internal and external data appended from time of the application Analytic tools and resources to develop the score Blind records or a hold out sample for validating the score A prudent implementation process Ongoing monitoring of score results What are some important features in the technology stack for supporting analytics? Flexible Fast Internal database “connectivity” Multiple underwriting configurations Champion challenger capabilities Easily connects to external data sources Easy data import-export capabilities Score development tools Feedback loops Questions for the panel? Thank you!