Winback Explorer

Welcome

In this module, you will learn about the ODP Reports  Winback Explorer, what you use it for and how to use it.

After completing this module, you should be able to:

• State the purpose of the Winback explorer
• Identify customers who are getting ready to churn
• Create a winback campaign
• Observe winback model from the data scientist viewpoint

What is it? / Why use it?

A customer's patterns of engagement with your brand determine their likelihood to remain your customer.

ODP's customer winback model helps identify customers who are deviating from normal engagement patterns, thus preventing churn. The model's primary inputs are "event days," marking active engagement as a "hit" and no engagement as a "miss." The more data sources integrated into ODP, the better the model understands a brand's engagement patterns.

The model scores customers based on their event patterns to predict their likelihood of remaining a customer and assigns them to one of three zones: Engaged, Winback, or Churned.

Customer engagement trend

This plot illustrates the relationship between the time since the last engagement and the likelihood of a customer remaining active. It shows a "cone" where the probability of remaining a customer drops differently for less-engaged (slower drop, upper border) versus frequently engaged (faster drop, lower border) customers.

The cone also includes areas denoted as Engaged, Winback, and Churned, specific to your brand's data.

• The Winback area begins where an increase in a customer's likelihood to remain a customer yields a correspondingly larger increase in their likelihood to re-engage in the next year. Start a winback campaign at this point.
• The Winback area ends where an increase in a customer's likelihood to remain a customer yields a correspondingly smaller increase in their likelihood to re-engage in the next year, meaning the customer can be considered churned.

Distribution of customer likelihood

A histogram displays the distribution of known customers by their likelihood to remain a customer (excluding anonymous users). This helps in understanding the audience composition. For example, a newsletter audience skewing Churned could affect open rates, or high-value customers slipping into Winback might require unique engagement strategies.

Customer winback model for Data Scientists

The model is based on the Pareto NBD (negative binomial distribution) model, a widely accepted customer lifetime value model, adapted to predict any kind of engagement rather than just purchase events. ODP validated that model is predictive of future interaction by making historical predictions and then looking at subsequent re-engagement.

The winback opportunity starts at the measure of liveness where increasing liveness has a larger corresponding increase on future engagement; this starting point is different for each client. The winback opportunity ends and a customer is considered churned when increasing liveness has a smaller corresponding increase on future engagement. Each section (Engaged, Winback, and Churned) has an aggregate likelihood of re-engagement with all activity held constant.

For more details on how to use Winback explorer properly, go through this demo: