Customer Lifetime Value: Count the Costs and Compare Equal-Age Cohorts
Calculate customer value with explicit cost assumptions, consistent cohort membership and equal observation periods, then separate forecasts from actual payback.
The Short Answer
Customer lifetime value needs a named horizon, customer population and cost definition. Start with observed contribution: retained customer revenue less the included product, fulfillment, payment, service and retention costs, divided by the original qualifying cohort count.
Compare cohorts at the same age, retain non-repeat customers and show future periods as unobserved. Keep forecast assumptions separate. Compare contribution before acquisition with the matching acquisition allocation and its timing, while distinguishing contribution payback from actual cash recovery.
Key takeaways
- A revenue-based LTV is not contribution after serving the customer.
- Keep the original acquired-customer denominator and compare equal ages.
- Separate observed value, allocated costs and future assumptions.
- Show timing and cost boundaries before using LTV to justify acquisition spending.
“Our customer lifetime value is $300” sounds useful until you ask what it means. Is that revenue already collected, a forecast, gross profit or money left after serving the customer? Was it measured over three months or three years? Does the average include customers who never returned?
The number can support a good decision only when those questions have answers. Revenue from repeat orders does not all become money available for acquisition. Products, delivery, payments, returns, service and retention activity still have costs.
I would begin with observed customer contribution over a defined period, then keep future assumptions separate. That produces a value measure the team can explain, compare and use without pretending it knows a customer's entire future.
Decide what kind of value the report describes
Separate three views. Lifetime revenue describes recorded sales under the system's definition. Observed contribution subtracts the included costs over the period you can actually observe. Forecast contribution adds assumptions about what may happen later.
Name the horizon and cost boundary beside the number. “Observed 180-day contribution per acquired customer, before acquisition cost” tells the reader much more than “LTV.” It also avoids suggesting that a six-month report covers the customer's complete lifetime.
Check platform definitions before borrowing their labels. Google's Analytics metric reference defines its LTV metric using purchase-related revenue events less refunds. That is a revenue measure; it does not establish the contribution after your product, fulfillment and service costs.
For a subscription business, distinguish the subscription relationship from the customer relationship. One person may cancel a plan and later make an ordinary purchase. Decide whether the report follows the subscription, the person across permitted linked orders or another defined relationship, then maintain that scope.
Define the customer group and keep its starting count
A cohort groups customers by a shared starting condition, often their first qualifying purchase period. Record the rule, customer identifiers, source systems and exclusions. Make sure repeat orders belong to the existing customer rather than creating a new acquired customer each time.
Use the original qualifying cohort count when calculating value per acquired customer. Someone who does not reorder is still part of the group you acquired. Removing inactive or refunded customers from the denominator can make the average appear healthier while hiding acquisition outcomes.
Document how test records, fraud, cancellations and unavailable history are handled. There may be a valid adjusted view, but show its definition and the original view rather than silently rewriting the group. A refund changes the economics even when it does not change cohort membership.
As a platform example, Shopify's customer cohort documentation describes grouping customers by first-order date and comparing their subsequent behavior. For a broader business, check whether your source includes all relevant channels and whether identity links are dependable and permitted.
Keep unlinked or uncertain records visible. Do not combine people merely because their names are similar, and do not assume a new device is a new customer. The value calculation needs an appropriate customer history, not just a website visitor count.
Build a cost definition the team can reproduce
Choose a management-reporting definition with the people responsible for the financial and operational records. The purpose here is to support decisions; formal accounting treatment still needs the business's accounting policy.
| Component | Decision to document | Common source of overstatement |
|---|---|---|
| Net customer revenue | Discounts, refunds, shipping revenue and tax treatment | Treating collected tax or later-refunded sales as available contribution |
| Product cost | Relevant sold-product costs and return adjustments | Missing costs for particular variants or counting returned units inconsistently |
| Fulfillment and delivery | Handling, packaging, outbound delivery and applicable return work | Looking only at the carrier's base charge |
| Payment and service | Fees and explicitly included variable support costs | Assuming every refunded fee is recovered |
| Retention activity | Included messaging, incentives and program costs | Deducting the same discount twice or leaving servicing costs out |
| Acquisition | Whether it sits outside or inside the displayed measure | Comparing with CAC after already subtracting that same CAC |
Use actual costs where available and show estimated allocations where necessary. A fixed monthly support bill may need a separate view from variable service cost. Do not label every allocation as an order-level observation.
Handle returns consistently. If revenue is already net of a refund, do not subtract the refunded selling price again. Record the product-cost and handling treatment for restocked, damaged or otherwise unrecoverable goods according to the chosen definition. Keep delayed adjustments visible.
The contribution-margin guide develops this boundary in more detail. The useful result is a reproducible bridge from orders to contribution, with unknown costs labeled rather than treated as zero.
Calculate one cohort from start to finish
Consider a fictional group of 1,000 acquired customers, each observed for 180 days after the first qualifying purchase. Assume one currency, no unlinked orders and the following simplified totals. These figures demonstrate the calculation; they are not benchmarks or a client result.
| Item through day 180 | Illustrative amount |
|---|---|
| Net revenue after discounts and refunds, excluding collected taxes | $120,000 |
| Product cost under the stated return treatment | $45,000 |
| Fulfillment, delivery and included return handling | $15,000 |
| Payment fees and included variable service cost | $6,000 |
| Included retention activity | $4,000 |
| Contribution before acquisition cost | $50,000 |
The contribution is $120,000 minus $70,000 of included costs. Dividing by the original 1,000 customers gives $50 of observed 180-day contribution per acquired customer. The corresponding revenue average is $120, which answers a different question.
Suppose the documented acquisition allocation is $38,000, or $38 per customer. Contribution after that acquisition cost is $12,000 in total, or $12 per original customer. It is not the company's net profit: overhead, financing, tax and other excluded items remain outside this simplified view.
If an omitted $5,000 of relevant servicing cost is discovered, contribution before acquisition falls to $45 per customer. The difference matters even though the sales history has not changed. Keep the revision and its reason in the report instead of overwriting the earlier number without explanation.
Compare customers at the same age
A cohort that began a year ago has had more time to buy again than one that began last month. Comparing their current cumulative totals can make the older group look better simply because it has existed longer.
Choose comparable ages such as 30, 90 or 180 days since each customer's starting event. Be explicit about whether the report uses exact elapsed days or calendar-month buckets. Those approaches can produce different exposure near month boundaries.
Do not fill unobserved periods with zero. Mark them unavailable until the group has had the required observation time, or use a clearly labeled partial view. If you calculate only for customers who have reached a given age, show that eligible population and avoid silently treating it as the whole newer cohort.
Check the mix as well as age. An acquisition promotion, product range change, new market or subscription offer can create a different group. Separate the segments that matter to the decision without cutting the sample into so many small pieces that random variation dominates.
Keep refunds and late costs on a consistent basis. You may need a current restated view and a record of what was known at an earlier decision date. Otherwise, apparent cohort differences can partly reflect how long each group has had for adjustments to arrive.
Keep forecasts visibly separate from observed value
Forecasting can help with planning, but future orders, retention and margins remain assumptions. Start with observed behavior from comparable, sufficiently mature groups and explain why it should or should not transfer to the new one.
For a subscription, do not automatically divide a recent monthly margin by one month's cancellation rate and present the result as certain lifetime value. The relationship can change with cohort age, pauses, payment failures, product fit and the definition of cancellation. A convenient formula can conceal those differences.
For occasional purchases, identify a credible reason and time for repeat behavior. Durable products, replenishable goods and apps with recurring fees do not share one universal repeat pattern. The subscription-retention guide helps separate customer and payment-related reasons when a recurring relationship is involved.
Use a small set of explicit scenarios: repeat probability, order contribution, timing and additional service cost. Show the observed portion separately from the forecast extension. Save the assumptions as they were when the decision was made and compare them later with actual outcomes.
For long horizons, discuss the appropriate treatment of future cash and uncertainty with the financial owner. Do not imply that a distant forecast dollar has the same decision value as money already available to fund the next order.
Put the timing of acquisition recovery beside the value
A customer group can eventually contribute more than its acquisition cost while still putting pressure on current cash. Show cumulative contribution and the assigned acquisition cost over time, using matching cohort membership and cost definitions.
In a separate fictional example, suppose contribution before acquisition reaches $25 per original customer by day 30, $33 by day 60 and $39 by day 90. With $38 of assigned acquisition cost, the measured contribution has crossed that amount by day 90. These checkpoints do not establish the exact day of crossing.
This is contribution payback under the definition, not proof of actual bank-cash recovery. Inventory payments, supplier terms, processor settlements, refunds and other cash movements can occur on different dates. Review that timing with the cash plan.
Avoid a universal acceptable LTV-to-CAC ratio. A ratio hides the observation horizon, remaining expenses, uncertainty and time needed to recover spending. The same ratio can describe businesses with very different operating risks.
Use the attribution reconciliation guide when channel cost and order records disagree. An allocated acquisition cost can be useful, but it should not masquerade as a perfectly observed causal path for every person.
Use the difference to choose a business action
When contribution is weak, identify what actually changed. Lower retained revenue, expensive delivery, high returns, costly support and fewer repeat orders call for different work. Do not respond to every weak value figure by sending more promotional messages.
For example, a customer group with frequent repeat orders may still contribute poorly if it needs deep discounts and expensive small shipments. A less frequent group may be healthier after costs. Check the actual driver before rewarding the behavior with another incentive.
Separate improvement in the whole cohort from selection effects. Reporting only customers who stayed active or joined a loyalty program can make a program look effective even if those people were already more likely to buy. Use an appropriate comparison when judging whether an intervention created additional value.
Give the next action an owner, an expected mechanism and an observation period suited to the purchase cycle. A clearer product description might reduce avoidable returns before it changes repeat purchase. An improved subscription control might reduce support work without increasing order frequency.
The weekly commercial review guide helps keep those decisions connected to the wider business instead of treating LTV as an isolated dashboard target.
Maintain the calculation as the business changes
Keep a definition sheet, source extracts or reproducible queries, allocation rules, observation cutoff and a log of corrections. Confirm that totals reconcile with the chosen business records before interpreting finer segments.
When a platform, product mix or cost arrangement changes, assess whether past comparisons remain valid. If the definition needs revision, explain the before-and-after boundary and restate comparable periods where practical. Do not hide a definition change inside an apparently improving trend.
Review who uses the number and for which decisions. Acquisition planning, retention work and overall profitability may need related but distinct views. A single unlabeled lifetime-value figure cannot reliably serve all three.
My recommendation is to prefer an understandable observed value over an impressive forecast. Once the costs, cohort and timing are clear, a model can add useful possibilities. Without that foundation, greater numerical precision only makes the uncertainty harder to see.
Questions about customer lifetime value
Should customers who never purchase again stay in the calculation?
Yes, when calculating value per acquired customer for the original cohort. Their first order, costs and lack of repeat activity are part of the outcome. You can separately analyze repeat buyers, but label that different population clearly.
Can the store's revenue-based LTV be compared directly with CAC?
It can describe revenue relative to acquisition spending, but it does not establish profitability or contribution payback. Deduct the relevant costs under a clear definition, align the customer group and horizon, and avoid subtracting acquisition cost twice.
What if we have only a few months of customer history?
Report the observed period accurately and keep longer-term forecasts separate. Use cautious, explicit scenarios if planning requires them. Do not turn missing future observations into zero, or assume repeat purchases merely to produce a lifetime number.
Explore the analytics and measurement guides for related work on customer definitions, order reconciliation and decisions based on dependable business records.
Sources
Still have a question?Ask the DTCUP Community ↗

