Segmenting LTV for Efficiency

Lifetime Value (LTV) is an important metric in acquisition efforts. We obviously don’t want to spend more acquiring a customer than they’re worth (LTV), so we use LTV as a benchmark to set our Cost per Acquisition (CPA) targets. In it’s simplest form, we’d assess the average LTV of our current customers to find an aggregate LTV, and from there apply a margin to determine the threshold CPA we’re willing to spend to acquire.

The Problem with Aggregate LTV

Utilizing an aggregate LTV can hinder our efficiency though, stopping us from acquiring higher valued and more expensive customers, while also fooling us into acquiring lower value customers that may be below our target CPA but aren’t worth what we’re spending to acquire them. 

Take a look at the slide to the right. 

In this example we’re assuming an all up average LTV of $80, and from there we are setting a CPS (Cost per Subscription) of $40. If $40 is our acceptable acquisition threshold – we’re missing out on subs who may cost $60 to acquire, but have a much higher LTV of $120. Alternatively, we’re acquiring subs who are under our CPS but may not be worth what we’re paying to acquire them. In the example to the right there’s a sub we acquired for $25 (well below our target $40 CPS), but only turned out to have an LTV of $20. Which means we lost money on that customer.

Breaking LTV into Segments

Here’s a different LTV tactic to take to minimize this problem. Instead of utilizing an all-up average LTV, assess the LTV of all your current customers and break them into segments. In the example to the right, instead of using an all up average LTV of $40, we’re breaking our customer base into three segments: $100, $60 and $30. From there we set individual CPS targets for each segment: $60, $30 and $10.

Once we have these three customer LTV segments, we can feed each one as a seed audience into media platforms to build LAL models off of – setting individual CPS targets for each segment. 

By doing so we can set higher thresholds for the more valuable customers (based on LALs), and set lower thresholds to stop us from spending too much to acquire the less valuable LTV customers. 

Of course this is just the beginning. From here we can build our own LAL models (instead of relying on ad platform models), we can add in propensity models for each segment to become more granular in our bidding and we can adjust our bid strategies within platforms to make sure we are going after the most valuable impressions and reducing wasteful spending on less-valuable impressions. But those are all topics for future posts. 

In the meantime, are you using an aggregate LTV for your CPA targets? Consider segmenting your LTV seeds into three groupings – it’s a fairly simple step to take towards improving acquisition efficiency.