Customer lifetime value for SaaS: why a static number isn't enough
A product leader justifies a feature investment on last quarter's LTV number, and three months later half that segment has quietly churned anyway.
A product leader pulls up the customer lifetime value figure the finance team calculated last quarter and uses it to justify a new feature investment aimed at the company's highest-value segment. Three months later, half of that segment has churned anyway, quietly, with no warning the static LTV number was ever capable of giving. The number wasn't wrong when it was calculated. It was just already out of date, because LTV built from historical averages describes where a cohort has been, not where it's headed.
Quick answer
Customer lifetime value (LTV) for a SaaS business is typically calculated as average revenue per account (ARPA) multiplied by gross margin, divided by customer churn rate, estimating the total revenue an average customer generates before they cancel.[1] The basic formula assumes churn happens at a steady rate across every account, which is rarely true. Feeding account-level churn risk scores into the calculation, generated by a churn-prediction model rather than a single blended churn rate, turns LTV into a forward-looking, segment-aware number instead of a historical average that quietly misrepresents which customers are actually worth investing in.
What LTV is and why the standard formula falls short
The standard SaaS LTV formula is simple by design: ARPA multiplied by gross margin, divided by customer churn rate.[1] It is useful as a planning baseline and it is also, by its own construction, an approximation that assumes every customer in a cohort churns at roughly the same rate over time.
Real churn does not behave that way. Most cohorts lose customers fastest in the first few months after signup, then the rate slows and stabilizes for the accounts that stick around. A formula built on a single blended churn rate smooths over that curve, which means it systematically overstates the value of newer cohorts and can understate the value of accounts that have already proven durable. The gap between prediction and reality is not a rounding error: a 2026 analysis of 35,512 customer cohorts found that 28.3% of cohorts saw actual revenue differ from the LTV estimate made at signup by more than 50%, and the direction of that error flips over time, with LTV tending to overestimate revenue at 12 months out but underestimate it by 24 and 36 months.[1]
That matters for a product or growth leader specifically because LTV usually drives real decisions: which segment gets the next feature investment, which channel gets more acquisition budget, which accounts get proactive customer success attention. A static, blended LTV number treats every account in a segment as equally likely to stick around, when in practice some accounts are already showing early signs of disengagement and others are showing signs of expansion. Churn prediction exists specifically to separate those two groups before the blended average hides the difference.
How LTV is normally calculated
The inputs are familiar to anyone who has touched SaaS metrics: ARPA, the average recurring revenue a single account generates; gross margin, the share of revenue left after the direct cost of serving a customer; and customer churn rate, the pace at which accounts cancel. Multiply the first two and divide by the third, and the result is an estimate of total revenue an average customer will generate before they leave.
A more advanced formula, developed by investor David Skok, adds a discount rate and an expansion-rate term to account for the fact that money received later is worth less than money received now, and that customers who stay often spend more over time rather than staying flat.[1] Both versions are still working from averages. Neither version, on its own, tells you which specific accounts are driving the number up or down.
Why a blended churn rate hides more than it reveals
A single company-wide churn rate assumes homogeneity that doesn't exist in any real customer base. An account that onboarded successfully, uses the product daily, and has never opened a support ticket about cancellation has a meaningfully different probability of churning than one that stopped logging in six weeks ago and missed its last two support responses. Both accounts get folded into the same churn rate in the basic LTV formula, which means both get assigned the same implied lifetime value, even though one of them is actively at risk and the other isn't.
Feeding churn-risk scores into LTV
This is where AI-powered churn prediction changes what LTV can actually do. Rather than applying one blended churn rate across an entire segment, a churn-prediction model scores individual accounts based on usage patterns, support interactions, billing events, and engagement trends, producing an account-level risk estimate rather than a single average.[2] Substituting that account-level risk score into the LTV calculation, in place of the blended churn rate, turns LTV from a static historical artifact into something closer to a live forecast: accounts showing early disengagement signals get a lower implied LTV immediately, well before they show up as a cancellation in next quarter's churn report.
What this changes for a product or growth leader
The practical shift is in what LTV gets used for. A static, segment-wide LTV number is suited to broad resourcing decisions made once a quarter. A risk-adjusted, account-level LTV supports faster, narrower decisions: which specific accounts should get a retention outreach this week, which upgrade path should be offered to an account showing expansion signals, and which low-risk, high-value accounts deserve the next round of feature-gating or pricing experiments without worrying about accelerating churn in that group.
Where this fits inside expansion and retention planning
LTV adjusted with churn-risk scoring also sharpens how a product leader reads expansion revenue. A segment with rising ARPA but a growing share of high-risk accounts is a less durable growth story than the aggregate expansion number suggests, and that distinction only becomes visible once churn risk is tracked at the account level rather than folded into a single rate for the whole segment.
A quick comparison: static LTV vs risk-adjusted LTV
| Static LTV (standard formula) | Risk-adjusted LTV | |
|---|---|---|
| Churn input | Single blended rate across a segment or company | Account-level risk score from a churn-prediction model |
| Update frequency | Typically recalculated quarterly | Updates as usage, billing, and support signals change |
| Best used for | High-level planning, board reporting | Prioritizing retention outreach, pricing, and expansion offers |
| Main weakness | Assumes linear, uniform churn across all accounts | Requires reliable usage and engagement data to score accurately |
Questions to ask before trusting an LTV number
- Is the churn rate feeding this LTV calculation a single blended figure, or does it reflect account-level risk?
- Has this LTV number been checked against actual cohort outcomes from a year or two ago, to see how far off the original estimate was?
- Is LTV segmented by acquisition channel or customer tier, or is it one company-wide average?
- Are the accounts with the highest implied LTV also the ones showing healthy usage and engagement signals, or is the number assuming stability that isn't there?
- Is LTV being recalculated often enough to reflect recent churn-risk signals, or only once a quarter?
Why this matters beyond one metric
None of this is really an argument against the LTV formula itself, which remains a useful starting point. It's an argument for not stopping there. A product or growth leader who treats LTV as a static, backward-looking average will keep making resourcing decisions based on a number that was already stale by the time it was calculated. Feeding churn-risk scores into that same formula doesn't require a new metric, just a better input, and it turns LTV into a number that moves with the business instead of describing where the business used to be.
Frequently asked questions
What is customer lifetime value (LTV) in SaaS? LTV is an estimate of the total revenue an average customer generates over their relationship with a subscription business, from signup through cancellation. The standard formula is ARPA multiplied by gross margin, divided by customer churn rate.
How do you calculate LTV for a SaaS company? Multiply average revenue per account (ARPA) by gross margin, then divide by customer churn rate. This basic formula is a useful planning baseline, though it assumes churn happens at a steady rate, which tends to overstate or understate true lifetime value depending on how far out the cohort is measured.
Why is the basic LTV formula inaccurate? It assumes linear churn across an entire cohort or segment, when real churn is typically front-loaded in the first few months and then stabilizes. Research on this pattern found over a quarter of cohorts saw actual revenue differ from the original LTV estimate by more than 50%.
What does churn prediction add to an LTV calculation? Churn-prediction models generate account-level risk scores based on usage, support, and billing signals. Substituting that risk score for a single blended churn rate turns LTV from a historical average into a forward-looking, account-specific estimate.
What is a good LTV to CAC ratio? A commonly cited rule of thumb is that LTV should be at least three times customer acquisition cost (CAC). This threshold is a starting benchmark rather than a guarantee of a healthy business, since it still depends on how accurately the underlying LTV number reflects real churn risk.
How often should LTV be recalculated? Standard practice recalculates LTV quarterly, but a risk-adjusted version built on account-level churn scores can update continuously as usage and engagement signals change, giving a more current read than a quarterly recalculation alone.
Can LTV be tracked at the individual account level instead of as a company average? Yes, when churn risk is scored per account rather than applied as one blended rate. This is what allows LTV to inform specific retention, pricing, and expansion decisions rather than only broad, segment-wide planning.
Sources
- ChartMogul, "Customer Lifetime Value (LTV)," updated September 8, 2026, and "LTV Has a Blind Spot" report (analysis of 35,512 customer cohorts, 2026). https://chartmogul.com/saas-metrics/ltv/
- ChartMogul, "Average Revenue Per Account (ARPA)," updated September 8, 2026 (churn prediction and segment tracking). https://chartmogul.com/saas-metrics/arpa/
Frequently asked questions
What is customer lifetime value (LTV) in SaaS?
LTV is an estimate of the total revenue an average customer generates over their relationship with a subscription business, from signup through cancellation. The standard formula is ARPA multiplied by gross margin, divided by customer churn rate.
How do you calculate LTV for a SaaS company?
Multiply average revenue per account (ARPA) by gross margin, then divide by customer churn rate. This basic formula is a useful planning baseline, though it assumes churn happens at a steady rate, which tends to overstate or understate true lifetime value depending on how far out the cohort is measured.
Why is the basic LTV formula inaccurate?
It assumes linear churn across an entire cohort or segment, when real churn is typically front-loaded in the first few months and then stabilizes. Research on this pattern found over a quarter of cohorts saw actual revenue differ from the original LTV estimate by more than 50%.
What does churn prediction add to an LTV calculation?
Churn-prediction models generate account-level risk scores based on usage, support, and billing signals. Substituting that risk score for a single blended churn rate turns LTV from a historical average into a forward-looking, account-specific estimate.
What is a good LTV to CAC ratio?
A commonly cited rule of thumb is that LTV should be at least three times customer acquisition cost (CAC). This threshold is a starting benchmark rather than a guarantee of a healthy business, since it still depends on how accurately the underlying LTV number reflects real churn risk.
How often should LTV be recalculated?
Standard practice recalculates LTV quarterly, but a risk-adjusted version built on account-level churn scores can update continuously as usage and engagement signals change, giving a more current read than a quarterly recalculation alone.
Can LTV be tracked at the individual account level instead of as a company average?
Yes, when churn risk is scored per account rather than applied as one blended rate. This is what allows LTV to inform specific retention, pricing, and expansion decisions rather than only broad, segment-wide planning.
Samuel builds the systems that turn billing events into trustworthy revenue numbers. He writes about retention, pricing, and the operational details that decide whether the metrics hold up.