Your dashboard says a new customer is worth $180. Twelve months later, the customers acquired in that campaign have produced only $92 each. Is the campaign failing, is the forecast too optimistic, or have the customers simply not had enough time to return?
Customer lifetime value is useful because it looks beyond the first order. It is also a forecast. Before spending more based on LTV, compare the estimate with what a mature group of customers has actually done.
This reality check does not replace a lifetime model. It shows whether its early assumptions are holding up.
Start with one cohort and one fair window
Choose customers whose first completed order happened in the same month or quarter. That is the cohort. Give every person the same observation window—12 months works well for many product businesses with several possible reorders.
Do not compare customers acquired last month with customers acquired last year. The newer group has had fewer chances to buy again.
Use completed, retained orders. Remove canceled orders, include refunds and discounts consistently, and keep every acquired customer in the denominator, including people who never returned. Excluding one-time buyers would inflate customer value.
For products with a long replacement cycle, 12 months may be too short. Use a window that allows a reasonable reorder opportunity, then label it clearly.
Measure realized 12-month value
The foundational guide to customer lifetime value explains revenue- and margin-based formulas. For this audit, calculate what the cohort produced inside the fixed window.
Realized 12-month revenue per acquired customer = cohort net revenue ÷ customers acquired
A more decision-ready version uses contribution after variable costs:
Realized 12-month contribution per acquired customer = cohort contribution ÷ customers acquired
Define contribution consistently. It may subtract product materials, packaging, transaction fees, picking and packing, variable shipping support, and discounts. Do not quietly compare contribution-based actuals with a revenue-based forecast.
A worked product-business example
A candle brand acquired 200 first-time customers during September. Its forecast assumed:
- average completed order value of $50;
- 2.4 orders per customer during the first 12 months;
- contribution of 45% after variable product and order costs.
The forecast was $50 × 2.4 = $120 in revenue per customer and $54 in contribution.
After 12 months, the cohort produced 350 completed orders and $16,800 in net revenue. Contribution after the brand’s defined variable costs was $6,720.
The realized results are:
- $16,800 ÷ 200 = $84 revenue per acquired customer;
- $6,720 ÷ 200 = $33.60 contribution per acquired customer;
- 350 ÷ 200 = 1.75 completed orders per customer;
- $16,800 ÷ 350 = $48 average completed order value.
The revenue forecast missed by $36 per customer. Most of the gap came from purchase frequency, not basket size: order value was only $2 below plan, while customers placed 1.75 orders rather than 2.4.
That diagnosis points toward a different action than a vague conclusion that “LTV is down.”
Build a forecast-versus-realized table
Use one table for each mature cohort:
| Input | Forecast | Realized at 12 months | Difference to investigate |
|---|---|---|---|
| Customers acquired | 200 | 200 | None |
| Orders per customer | 2.40 | 1.75 | Reorder behavior or timing |
| Average completed order | $50 | $48 | Product mix, discounts, returns |
| Contribution rate | 45% | 40% | Costs, promotions, shipping support |
| Contribution per customer | $54 | $33.60 | Combined effect |
Preserve the forecast that existed when the decision was made. Replacing it later with actual results destroys the comparison.
Diagnose why the forecast missed
Repeat behavior
Did fewer customers return, or did returning customers buy less often? The guide to measuring repeat purchase rate by cohort helps separate those questions.
Order value
Did customers choose smaller products, use larger discounts, or avoid bundles? Review average order value without treating it as profit.
Margin and contribution
Supplier costs, packaging, payment fees, fulfillment labor, shipping support, and promotions may have changed after the forecast was built. A revenue forecast can look accurate while contribution falls short.
Refunds and cancellations
If the model used gross sales but the audit uses retained net sales, fix the definition. If both use net sales, investigate whether returns increased for a particular first product, campaign, or season.
Timing
Some purchases may occur after month 12. That does not make the audit wrong; it means 12-month value and full LTV answer different questions. Track both instead of extending the window until the result looks favorable.
Customer mix
A campaign may attract gift buyers, subscribers, wholesale prospects, or replenishment customers in different proportions. Compare segments only when each group is large and clean enough to support action.
Decide what to change
Use the variance to change one of four things:
- The forecast. Lower an unsupported purchase-frequency, order-value, margin, or lifespan assumption.
- The customer experience. Test clearer product education, replenishment timing, availability, or post-purchase support when the evidence points there.
- The acquisition limit. Recalculate what the business can afford to spend using a consistently defined contribution figure. The LTV-to-CAC guide can support that step.
- The measurement window. Change it only when the product’s natural purchase cycle makes the current window unfair, then apply the new definition consistently.
Do not promise that a retention tactic will increase LTV. Run a defined test and wait for eligible customers to have a fair opportunity to reorder.
Make the review repeatable
Once per quarter, close the next mature cohort and record:
- acquisition period and source;
- customers acquired;
- observation window;
- completed orders, net revenue, and contribution;
- refunds and discounts;
- forecast assumptions;
- realized results and variance;
- one decision, owner, and review date.
Several cohorts reveal more than one blended average. They show whether improvement is persistent or whether one launch, season, or promotion distorted the headline.
Frequently asked questions
Is 12-month customer value the same as lifetime value?
No. It is realized value inside a fixed window. Lifetime value estimates the full relationship, which may extend beyond 12 months.
Should customers who never reorder stay in the calculation?
Yes. If they were acquired into the cohort, they belong in the denominator. Removing them overstates average realized value.
Should LTV use revenue, gross profit, or contribution?
Any can be reported if it is labeled clearly. Contribution is often more useful for spending decisions because it reflects defined variable costs, but it still is not net profit.
How old should a customer cohort be before reviewing it?
Old enough for every customer to receive the full observation window and a reasonable reorder opportunity. A 12-month audit requires a cohort whose last member was acquired at least 12 months ago.
What if subscriptions and one-time buyers behave very differently?
Report them separately when the data is reliable. A blended average can hide both strong subscription retention and weak one-time reordering.
Practical takeaway
Pick one mature first-order cohort. Compare forecast order frequency, order value, and contribution with 12 months of completed behavior. Keep one-time buyers in the denominator and preserve the original forecast.
The gap is not a grade. It is a clue. Use it to correct one assumption or test one customer-experience change before increasing spending. Repeat customers matter, but the evidence has to arrive before their future value becomes money the business can count on.
