Every communication company wants to inoculate its portfolio against bad debt, late payments and painful collections. But many still use traditional generic risk models to uncover potential problems, either because they’ve always used generics or because they see their limited predictive abilities as adequate enough.
Generalization dilutes results
The main problem with generics, however, is how they generalize consumers’ payment behavior and delinquencies across credit cards, mortgages, auto loans and other products. They do not include payment and behavioral data focused on actual communications customers only. Moreover, their scoring methodologies can be too broad to provide the performance, lift or behavioral insights today’s providers strive to attain.
Advantages of industry-specific models
Communications-specific modeling can be more predictive, if you want to know who’s more likely to prioritize their phone bill and remit promptly, and who’s not. In multiple market validations, pitting an optimized industry-specific model against traditional generic products, Experian’s Tele-Risk ModelSM and Telecommunications, Energy and Cable (TEC) Risk ModelSM more accurately predicted the likelihood of future serious delinquent or derogatory payment behavior. Compared with generics, they also:
- Provided a stronger separation of good and bad accounts
- More precisely classified good vs. bad risk through improved rank ordering
- Accurately scored more consumers than a generic score that might have otherwise been considered unscorable
Anatomy of a risk score
These industry risk models are built and optimized using TEC-specific data elements and sample populations, which makes them measurably more predictive for evaluating new or existing communications customers. Optimization also helps identify other potentially troublesome segments, including those that might require special handling during on boarding, “turn ons,” or managing delinquency.
Check the vital signs
To assess the health of your portfolio, ask a few simple questions: Does your risk model reflect unique behaviors of actual communications customers? Is overly generic data suppressing lift and masking hidden risk? Could you score more files that are currently deemed unscorable?
Unless the answer is ‘yes’ to all, your model probably needs a check-up—stat.