Most teams calculate customer lifetime value with a formula that quietly hides the problem it is supposed to solve. Average revenue per customer times average customer lifespan produces a single, tidy number, and that number is close to useless for deciding where to actually invest, because it treats a customer who churns in month two the same as one who has been active for three years, just averaged into the same figure.
What Customer Lifetime Value Is Supposed to Tell You
Customer lifetime value is meant to answer a specific question: how much is a customer worth over the full length of their relationship with the business, so that acquisition spend, retention investment, and service levels can be sized sensibly against that number. Get CLV right and it becomes the number that justifies spending more to keep a customer than to acquire a new one. Get it wrong and it either overstates what customers are worth, which leads to overspending on acquisition, or understates it, which leads to under-investing in retention.
The simple formula, average order value multiplied by purchase frequency multiplied by average customer lifespan, is a reasonable starting point for a single blended number. The problem is that almost no customer base behaves like its average. A meaningful CLV model has to account for the fact that a small share of customers usually generate a disproportionate share of lifetime value, and that segment needs to be visible, not buried inside an average.
Historic vs Predictive Customer Lifetime Value
Historic CLV looks backward: it sums actual revenue from a customer to date. It is simple, accurate for what already happened, and useless for forward planning, because it says nothing about a customer who joined last month and has not had time to prove their pattern yet.
Predictive CLV estimates future value based on patterns observed in similar customers, using early signals like first-week engagement, first order size, or activation behaviour to project forward. This is the version that actually informs decisions like acquisition budget by channel or which segments deserve a heavier retention investment, because it gives you a number before the customer has fully proven themselves rather than after.
Calculating Customer Lifetime Value by Segment, Not in Aggregate
Step 1: Start with historic CLV by cohort. Group customers by acquisition month or acquisition channel and calculate actual revenue to date for each cohort. This alone usually reveals that some channels or campaigns are bringing in customers worth two to three times more than others, a difference a single blended CLV number would hide entirely.
Step 2: Layer in [RFM scoring](/writing/rfm-analysis-step-by-step) or a similar behavioural segment. Customers who score high on recency and frequency early in their lifecycle are the ones a predictive model should weight most heavily. This is where CLV stops being a finance exercise and starts connecting to the same segmentation work that drives lifecycle messaging.
Step 3: Build a churn-adjusted projection. Rather than assuming every customer stays until an arbitrary lifespan number, weight the projection by the actual survival curve for that cohort: what percentage are still active at month three, month six, month twelve. Subscription benchmark research is a useful sanity check here if your own historical data set is still small.
Step 4: Separate acquisition cost from the CLV number itself. CLV and customer acquisition cost are often reported together as a ratio, but conflating them in the calculation itself makes it harder to see which lever, acquisition efficiency or retention, is actually driving the ratio in a given period.
Step 5: Revisit the model quarterly. A CLV model built on last year's churn curve and this year's pricing will drift out of accuracy quietly. Treat it the same way you would treat any retention metric: reviewed against real outcomes on a schedule, not calculated once and left alone.
Why CLV and Retention Work Are the Same Problem Viewed Differently
CLV is often treated as a finance metric and retention as a marketing metric, run by different teams with different dashboards. In practice they are the same underlying question asked two ways: finance asks what a customer is worth, marketing asks how to keep that customer active. A retention programme that lifts D30 or D60 retention is, by definition, lifting CLV, and a CLV model that is not connected to the actual lifecycle programme driving retention is disconnected from the lever that would move it.
This is also why CLV by segment matters more than CLV in aggregate for making investment decisions. If the top quartile of customers by predicted CLV are not receiving meaningfully different treatment than the bottom quartile, the business is spending the same retention effort on customers worth very different amounts, which is rarely the right allocation.
It is also worth stating the underlying economics plainly, because the CLV model only matters if the comparison it is feeding is real. Acquiring a new customer consistently costs more than retaining an existing one, often by a wide margin, and CLV is the number that makes that comparison concrete rather than a slogan repeated in a strategy deck. A business that cannot state its own CLV by segment is, in effect, guessing at how much retention work is worth funding.
Common Mistakes When Calculating Customer Lifetime Value
Using a single blended number for every decision. Acquisition budget by channel, retention investment by segment, and support service levels all need CLV cut differently. One number used for all three usually gets at least one of them wrong.
Ignoring gross margin. Revenue-based CLV overstates value for lower-margin products or customer segments. A CLV model used to justify acquisition spend should be built on contribution margin, not top-line revenue, or the payback period will look better than it actually is. This is one of the more common gaps in SaaS metrics discussions, where CLV and CAC ratios get compared without checking that both sides of the ratio are measured the same way.
Treating the model as static. Pricing changes, product changes, and shifts in the customer base all move the real CLV number. A model calculated once during a fundraising deck and never updated is a historical artefact, not a working tool.
Building it in isolation from the retention programme. A finance team calculating CLV with no visibility into what lifecycle marketing is actually doing, and a marketing team running journeys with no visibility into which segments carry the most lifetime value, will optimise against different, sometimes conflicting, goals.
Customer lifetime value is only useful when it changes a decision: which channel gets more budget, which segment gets a better experience, which customers are worth a human touch instead of an automated one. A precise CLV number that does not feed any of those decisions is an exercise. A rough one that does is worth more.
Frequently Asked Questions
What is the simplest way to calculate customer lifetime value?
Multiply average order value by purchase frequency by average customer lifespan. This gives a single blended number that is a reasonable starting point but hides the difference between high-value and low-value segments, so it should not be the only version calculated.
What is the difference between historic and predictive CLV?
Historic CLV sums actual revenue a customer has generated to date. Predictive CLV estimates future value using early behavioural signals and patterns from similar customers, which makes it useful for decisions about customers who have not been active long enough to prove their pattern yet.
Should customer lifetime value include gross margin or just revenue?
For decisions involving spend, such as acquisition budget or retention investment, CLV should be built on contribution margin rather than revenue. Revenue-only CLV overstates value in lower-margin segments and can justify spending more than a customer relationship actually returns.
How often should a CLV model be recalculated?
Quarterly at minimum, and sooner after any pricing change, major product change, or shift in acquisition mix. A model built on an outdated churn curve or price point will quietly misrepresent what customers are actually worth.
How does CLV connect to retention marketing?
CLV and retention are the same underlying question viewed from finance and marketing respectively. Any lift in retention by definition lifts CLV, and a CLV model that is not connected to which segments the retention programme is actually improving will not reflect where the business should invest next.