Easy as it might seem, this formula can generate confusing results.
Why is it so challenging?
Let me give you a real-life example:
We’ve recently calculated the CLV for a company and… surprise surprise – we got 4 different values:
$162 $222 $814 $1333
That’s quite a range. Why such a huge discrepancy?
In our case, it turns out that the initial result we got, $1333, was unrealistic. Nobody believed it.
So we decided to clean the data out. We started with test orders because the company considered them negligible. We discovered they weren’t negligible at all. The second estimate was of $814. Still, this seemed far-fetched.
We also noticed they put yearly and monthly customers together. What we did next was to calculate churn and ARPA just for monthly customers.
Our third result was $162. We came to this result after excluding all the users whose subscriptions had expired that month. This result was far less than the initial calculation and seemed too bad to be true.
It turned out that a considerable proportion of users (roughly 10%) purchased another subscription after the first one had expired. So we included them in our fourth attempt. With the final numbers in place, the CLV for monthly users was $222.
We took our analysis one step further and calculated the CLV based on acquisition channels. Clearly, not all users are the same.
For the company in question, organic channels generated more than 60% of customers, who had a lifetime value of $255, while paid channels generated 40% of the customers, who had a lifetime value of $172.50.
This breakdown gave the company a much clearer view of what their cost of acquisition should be per channel.