Attrition Model: How does the attrition model work? скачать в хорошем качестве

Attrition Model: How does the attrition model work?

What we've seen with legacy attrition prediction models in the market is they often rely on slow moving data. For example the number of branch visits that a customer makes has historically been widely used in models that project churn. But how relevant is that really today? So we start with better data, data that is produced multiple times throughout the day and when used properly can be oftentimes very predictive in detecting churn. The key is transaction data, things like purchases or debit card swipes at retailers micro deposits made to competing financial institutions and even cash infusions to disrupters in the industry like chime and others. We like to answer questions, for example,h if we see a recurring payment to t-mobile come to an end on an account is it predictive of churn? Is that customer going to leave the institution or can we quickly and easily see just a shift in spend within the same category to another competitor in their industry like verizon? This is not a churn concern at all, and it's that enriched transaction data of finally making it relevant and useful in predictive models.