Onboarding is often thought of as an early-stage retention strategy. It's also an effective way to improve long-term customer value.
Setting up proper customer onboarding treatments, based on predicted payment behavior and predicted customer value, will minimize bad debt on the back-end. As an example, a top-five direct broadcast company implemented prepayment and auto-pay strategies during the sign-up process, based on out-of-the-box and custom credit worthiness models, as well as predicted subscriber profitability models. The result: A significant decrease in bad debt churn and write-offs on the backend.
The new onboarding process entices the subscriber to pay and become loyal based on two dimensions:
- Subscriber predicted future value
- Probability of payment over time
Those two dimensions, enabled by predictive models, are powerful if used within a proper treatment strategy. For example, one of the first questions a company should pose is, "Once the analytics are available, should we not allow subscribers to activate who have a high probability of not paying?"
Answer: Absolutely not! You do not want to shut off the spigot altogether. Instead, ask those potential customers to provide a prepayment, which may be applied to the ongoing monthly bill for a certain period of time so they are not paying more than they need to. Then show them goodwill, as well, by providing a prize at the end of the prepayment time period. Further, the prepayment period should be determined by the level of creditworthiness and the prize at the end of the rainbow should be determined by future predicted value.
The key to a proper customer onboarding process is the strategic application of the resulting treatments. The predictive models for the most part are straightforward; it's how the analytics is converted into proper customer treatments, and then tracked and optimized, that will provide the bottom line profit impact. Just as with collection efforts, the acquisition treatment strategy, and ongoing tracking, is the key that unlocks incremental profit.
Prospects must be segmented based on the models and other available data into unique treatment buckets so that the proper onboarding offer is made. The process, at a glance, is the following:
- Using previous prospect and customer history, develop a payment model at the point of acquisition. In the direct broadcast/cable world, the initial model typically predicts the probability of nonpayment in the first three to four months; however, that time period varies based on the industry. Customer survival curves will help you better understand how far out to predict, as new-customer nonpayment typically spikes at a specific period.
- Using previous prospect and customer history, develop a predicted customer value metric at the point of acquisition that may be applied to all prospects. This could be as complex as a true Life Time Value (LTV) metric or as simple as a profit calculation. Start simple, incorporate, and then enhance -- as something is always better than nothing. Remember, this is not a financial metric that must be reported to the SEC. It's a means to segment prospects.
- Develop additional prospect segmentations or models that are appropriate to your environment. Examples include response models and demographic-based segmentations, which may be used to tailor messaging or channel.
- Use all that was developed above along with other available data to bucket prospects into unique initial segments that will be assigned onboarding treatments. As an example, high probability to not pay, but high potential value.
- Using the segments developed in the previous step complete an analysis that uses historical data that will:
a. Determine nonpayment attrition rate and the average cost of an attritor when they attrite for each segment, as this is the cost that must be asked for upfront to eliminate your current exposure.
b. Based on the result above, set the onboarding treatment such that the initial cost to the customer outweighs the current aggregate exposure segment.
c. Use the predicted customer value to set the long-term incentive.
The process above explains how to setup the system; that's the easy part. Optimizing over the long term is the trick, as it requires the ability to be nimble and adjust your treatments on the fly. Some key considerations include
- Although this is not popular, a small percentage of prospects should flow through the onboarding process with no collections treatment. This is not popular as it is seen as leaving money on the table, but trust me you will be leaving money on the table without that group. That group may be used to determine the true effectiveness of the onboarding process, as well as to optimize long term, as those customers did not receive treatments that affect future behavior.
- Initially, start small with the number of treatment groups, learn, and then expand the groups. It is human nature to want to create many groups, as that is the point (to segment and set proper treatments at a detailed level); but starting small allows for true measurable results that you can use to build momentum in the organization.
- Once your process is well-oiled and many treatment groups are being used, more advanced analytic testing techniques (such as fractional factorial) may be used to draw inferences across and within groups that may otherwise be too small on their own.
The keys to a fine-tuned onboarding process, used to minimize write-offs on the backend, are predictive models, actionable treatments, and an ongoing test-and-learn process. And, the beauty in creating this process is that fancy and expansive tools are not needed in many environments. Industry standard database and analytic tools will work just fine and many available robust options are open source.
+ + + + + + +