An important use case for CLV is early indication of top customers beginning to disengage. This is especially meaningful for companies with the capacity to perform personalized outreach, or the technical maturity to perform automated outreach, to customers likely to leave.
Engaging a customer weeks or days before they leave is far likelier to succeed than trying to engage an already-churned customer.
But how can we know when a customer goes from considering their other options to completely disengaging? How would Chipotle know if I just need a break from their sofritas recipe versus deciding I'll be going to Qdoba from now on? Often we can't know for sure, but we can leverage tools like a CLV model to make much better-informed decisions using data.
Let me show you a real example for a customer of an eCommerce company. Keep in mind this is historical data - we're viewing this company as if we are in December of 2021.
This customer started out strong providing ~$300 in revenue on Aug 31st 2020. They return only a little more than a month later to give another ~$350 on Oct 3rd 2020.
The customer continues with this strong behavior, purchasing from this company every 1-2 months.
Come February, something concerning happens: it takes five months until we see this customer again. Interestingly, this is also the largest purchase of this customer's history. Maybe they just skipped a purchasing cycle and are now making it up? But the trend continues: as of December 2021, we haven't seen this customer again since their July purchase.
The important question is: does this reflect a new normal purchasing behavior for this customer considering their last purchase took 5 months? Or is this customer at risk of leaving? We can't know for sure.
Let's turn to a CLV model to give us concrete information to help us make a better decision about what to do.
When we look to our CLV model, we see that based on this customer's purchase history compared to their peers who started in the same cohort, there's a 79% chance the customer will return, with an expected customer lifetime value (revenue in the next 12 months in this case) of $354.
Considering the value this customer is providing to this company - an expected $354 in the next 12 months - it becomes easier to compare the cost of personalized outreach or a reactivation campaign to what we reasonably expect to receive from this customer in the future.
For larger companies, grouping many customers like this one and sending an enticing email (for example, with offers on similar products they may enjoy), and comparing to the baseline for other customers like this one who do not receive the email, could be a helpful way to learn more about these "on-the-fence" customers and what makes them excited to continue to engage with the company. Specifically, we can measure the change in probability to return, purchase frequency, and customer lifetime value of this campaign, and consider whether this was worth the cost.
Thinking about your company, given your current processes, how would you approach this customer with this particular transaction history?
Would you have flagged them in the first place?
Would you leave it up to them to decide whether to continue to engage with your company or would you try to conduct some form of outreach?
How would you measure whether this outreach is working? If not, what else would you try?
As your company's customer analytics practice matures, you'll find that the answer to one question becomes obvious, allowing you to focus on the next most valuable question.
Comments