Customer lifetime value: 7 Steps to Calculate and Use It Carefully

Customer lifetime value is an estimate of the value a customer contributes over the relationship with a business. Some reports measure lifetime revenue, while others measure gross profit or contribution after relevant costs. The definition matters because the same customer can look attractive in a revenue model and less attractive after delivery and support costs.

For trades and service businesses, repeat work often arrives unevenly. A maintenance customer may book regularly, while a replacement project customer may return years later. This article explains seven steps for a practical customer lifetime value calculation, with examples, cohort rules, and checks that keep the estimate useful.

What does customer lifetime value tell you?

Customer lifetime value helps you compare the long-term economics of customer groups. It can inform acquisition spending, service design, and retention priorities. However, it is an estimate, not money already in the bank.

Twilio’s overview of customer lifetime value describes several calculation approaches. Choose a method that fits your data and decision rather than adopting the most complex formula available.

Historical value measures what a customer has generated so far. Predicted value estimates what may happen later. Keep those measures separate. A new customer has little historical revenue, but that does not automatically mean they have low future potential.

Also distinguish value from cash timing. A profitable customer relationship may still create funding pressure if you pay delivery costs long before collecting invoices. Lifetime value should sit beside cash and payment measures, not replace them.

1. Define the customer and the value basis

First, decide what counts as one customer. It may be a household, site, company, or account group. If one property manager controls several sites, the right unit depends on whether you want site economics or account economics.

Next, define value. A revenue model adds expected sales. A gross-profit model subtracts cost of sales. A contribution model subtracts the relevant variable costs you choose for the decision. Write the definition beside the number.

Use a consistent treatment for discounts, refunds, credits, tax, and canceled jobs. For example, invoiced revenue and collected revenue answer different questions. Do not switch between them without explaining the change.

Set the time basis as well. A three-year value estimate is easier to interpret when you label it as three years rather than calling it an unlimited lifetime. A fixed horizon can work well when you lack enough evidence for a longer prediction.

2. Build a reliable customer history

Connect jobs, invoices, credits, and support records to a stable customer identifier. Otherwise, repeat work may appear as several new customers, which lowers measured frequency and distorts the relationship history.

Review duplicate accounts and changes in ownership. A business name change should not necessarily create a new customer, while a new homeowner at the same address may represent a different relationship.

Anomoz’s guide to clean CRM records covers the data structure behind this work. Reliable identity matching often matters more than a sophisticated prediction model.

Keep first-purchase date, transaction dates, net revenue, and the relevant cost fields. Record your data coverage. If the system contains only two years of history, some customers may have started their relationship much earlier.

Also distinguish missing history from inactivity. An account imported last month may look new even when staff have served it for a decade. Label uncertain start dates rather than treating them as complete evidence.

3. Start with a simple model you can explain

A basic revenue estimate uses average order value, purchase frequency, and expected relationship length. Use the same time unit for frequency and length.

Suppose a fictional customer group spends $250 per visit, books twice a year, and remains active for an estimated four years. The model gives $250 × 2 × 4 = $2,000 in estimated lifetime revenue.

This is not $2,000 in profit. If the relevant contribution margin is 35% under your cost definition, the rough contribution estimate is $700 before any other costs excluded from that margin.

Keep the assumptions visible. A simple model can help with an early decision, but it can hide changes in purchase frequency, margins, and survival over time. Use it as a starting point, not an exact forecast.

4. Use cohorts when customer behavior varies

A cohort groups customers by a shared starting point, such as the month of first purchase. Following cohorts lets you compare behavior at the same relationship age.

For example, compare the first twelve months of customers acquired through two channels. Do not compare a three-year-old group with customers who joined last month and conclude the older group has higher lifetime potential simply because it has had more time.

Track repeat purchases, net revenue, and contribution by elapsed period. Then examine whether later cohorts follow a similar pattern. Changes in pricing, service mix, or acquisition channels may make older behavior less relevant.

For retention definitions, see customer retention rate. A consistent cohort population helps prevent customers from appearing or disappearing solely because the report changed its rules.

Use enough history to support the time horizon. If no cohort has reached year four, a four-year estimate needs explicit assumptions. Show the observed portion separately from the projected portion.

5. Account for costs and service patterns

Customers with equal revenue may create different value. One may need standard visits near existing routes. Another may require distant travel, repeated rescheduling, or extensive account support.

Choose cost allocations that fit the decision. Avoid spreading every overhead line across customers without considering whether it changes the action. Equally, do not ignore material delivery costs because they are hard to assign.

Anomoz’s article on cost to serve explains how service demands can affect customer economics. Use that view alongside revenue when planning offers or service levels.

Consider the natural purchase cycle. A roof replacement customer may remain satisfied for many years without another large purchase. A simple “no purchase in twelve months” rule could wrongly label that customer as lost.

Separate recurring plans, repeat repairs, and occasional projects where their behavior differs. A model built for monthly subscriptions may not fit an irregular field-service relationship.

6. Compare value with acquisition cost and payback

Lifetime value becomes more useful when you compare it with the cost of acquiring the same customer population. Use compatible definitions. A lifetime contribution estimate should not be compared with only advertising clicks if your acquisition process also requires paid sales time.

Salesforce’s customer acquisition overview explains acquisition channels and cost. For your own comparison, align the customer definition, channel treatment, and time window before interpreting a ratio.

In the fictional example, a $700 lifetime contribution estimate and a $200 acquisition cost leave $500 before costs that the model excludes. However, that does not mean the customer creates $500 in immediate cash or final net profit.

Check payback too. If contribution arrives slowly, a seemingly attractive lifetime ratio can still create a cash burden. Model when the customer generates contribution and when the business pays acquisition and service costs.

Avoid treating any single ratio as a universal target. Different service cycles, uncertainty, margins, and funding constraints call for different decisions. Test assumptions against your own operating history.

7. Test decisions and refresh the model

Use the estimate to compare specific actions. For example, should you invest more in a channel whose customers repeat more often, or improve onboarding for a group that stops booking early?

Start with a limited change and measure results. Do not assume a high-value group will respond to every offer. A discount may increase bookings while reducing contribution, so track both effects.

Review the model against actual cohort outcomes. If projected repeat visits fail to appear, adjust the assumption. Keep a version history so people can explain why the value estimate changed.

Also monitor data drift. New services, pricing, service areas, or customer mix can make old averages less useful. Assign a model owner and a review schedule rather than letting a spreadsheet become a permanent fact.

For connected customer histories and follow-up rules, custom crm development may support the process. Define the data and decisions first, then assess whether the current tools already meet the need.

A cohort example with a fixed time horizon

Consider 100 fictional customers who join in the same quarter. During their first year, they generate $40,000 in net revenue. Under the company’s chosen cost definition, they generate $14,000 in contribution.

The observed first-year contribution per acquired customer is $14,000 ÷ 100 = $140. Keep the original 100 as the denominator when reporting value per acquired customer, even if only 60 buy again.

If the business divides later revenue only by active repeat customers, it answers a different question: value per active customer in that period. That can help service planning, but it overstates value per acquired customer if you use it without adjustment.

Next, the business may forecast year two using repeat-purchase evidence from comparable older cohorts. Label that portion as projected. Add a lower-repeat scenario to show how uncertainty changes the result.

Common errors that create false confidence

Mixing time units is a frequent mistake. Monthly purchase frequency multiplied by a relationship length in years produces the wrong result unless you convert the units.

Another error is treating one divided by churn as a reliable lifetime for every business. That shortcut depends on strong assumptions about stable churn and the relationship model. It can mislead when behavior changes with tenure or purchases happen irregularly.

Do not use a long historical average without checking the current service mix. An old cohort may have bought a different plan at a different price. Also avoid reporting more decimal places than the inputs justify.

Finally, do not let a low predicted value override basic service commitments. Models can be wrong, and customers can change their behavior. Use estimates to support decisions with suitable review, not to create unexplained barriers.

Show a range when the future is uncertain

Test the assumptions that have the largest effect on value. For a service plan, that may be repeat visits and the margin per visit. Keep the other inputs fixed while changing one assumption so the team can see its effect.

Then build a few plausible combined scenarios. Use your own records to support the range where possible. Do not present the best case as the expected case. A decision that still works under a cautious scenario may need less confidence in distant repeat sales.

Use value estimates as a decision tool

Customer lifetime value helps when the customer unit, cost basis, and forecast horizon remain clear. Start with observed data, separate projections, and check the model against real outcomes. Then use it to test a specific acquisition, service, or retention decision.