Table of contents for Live from InterACT 2008
- Live from InterACT – Ian Ayres
- Live from InterACT – New Approaches to Strategies
- Live from InterACT – Insurance in the 21st Century
- Live from InterACT – Building a Decision Engine
- Live from InterACT – Using Risk Applications to Drive Growth
- Live from InterACT – Design for People, Build for Change
- Live from InterACT – Automate, Improve and Connect
- Live from InterACT – The Mortgage Crisis
- Live from InterACT – Scoring Innovation
- Live from InterACT – Optimal Pricing
- Live from InterACT – An Enterprise Decision Engine for Originations
- Live from InterACT – Changing the game
- Live from InterACT – Closing Keynote
A collections session next with folks from Adeptra, Fair Isaac and GE Money talking about GE’s vision for virtual collections. The collections environment is extremely bad this year with massive growth in the need for collection agents. Delinquencies are up, problems are up, consumers are stressed. Scores are worsening (credit profiles are worse), payments are getting smaller, it’s becoming harder to collect at a given score and the rate of high risk accounts is growing by 16%.
The declining rates of closure, combined with more delinquencies, has led to 100% growth in staffing at GE Money. However the rate of contacts is dropping fast – consumers are not responding to phone calls – so a key focus is on broadening the range of channels that can be used. Besides new channels and a focus on delinquencies, GE Money is also aggressively implementing new approaches and using champion/challenger to test them rigorously given the difficult environment. In particular, using segmentation and champion/challenger testing to find the right strategy and to optimize multiple contacts.
The old approach would be more people with better negotiation skills but the call rates are so bad this seems marginal. More contacts with a consumer would usually be considered better but adding channels might be done piecemeal and some automation might be deployed but this becomes confusing to consumers and makes it hard to do analytics.
Adeptra’s solution allows multiple channels – text, phone, mail, email, web – to be woven together with rules and analytics to deliver targeted treatments to different kinds of consumers. This approach is much cheaper than using a human collector and yet, for some target segments, as effective as human collectors. Adeptra’s solution is two-way, personalized and rules-driven. It is designed to give consumers choices and options to engage them more. The different channels are integrated with each other, with other core systems and with other sources of information. The actions are taken in real-time and mostly automatically, typically engaging the consumer in the process e.g. by allowing them to select the web as a channel. Rules handling things like whether an email was opened, whether a consumer has gone to the web and what the consequences in terms of calls or other interactions should be. Each treatment is unique, targeted to that consumer etc – what I have called extreme personalization – yet coordinated so that the same offers show up on different channels.
It helps GE Money keep up with changing consumer preferences and make it a marketing opportunity that fits the way the consumer thinks. GE’s approach to this – virtual collections -requires a number of things:
- Real-time architecture to manage the whole set of options as the calls evolve
- Discrete, personalized, architecture – case by case
- Integrates analytics – which accounts to go virtual, best treatment
- Integrated rules – the next appropriate step, looking for the best resolution
GE Money has found that there are definitely segments that can be resolved effectively by this virtual approach and that analytics can help identify these segments. Segmentation and rules-based routing remain very important and, despite aiming for auto-resolution, early staff contact with consumers is still really important.
There is a lot of analytics in use in the process. GE Money has some 48 models over 30 portfolios that predict risk of delinquency at different stages. To keep up with changes in the collection environment and portfolio mix, models are retrained every 3-6 months. The use and updating of these scores is very dynamic and a direct feedback loop helps keep them up to date. Feedback on what was tried, how it worked etc is key to improving the models. Retrained models show a particularly large improvement when used in the very overdue accounts, the worst problem, though they are better than older models at every stage. This higher accuracy helps improve segmentation (traditional collections, self correct or virtual agent) and to predict the best work-out plan and settlement amount.
Overall the virtual agents help manage the current difficult times while reducing costs and improving performance with automated resolution.