The customer lifetime value on the board slide belongs to a customer who does not exist.
It is an average, and an average blends people with little in common. It adds the member who will stay for a decade to the member who will leave at her first renewal. Then it reports them as one customer nobody will ever serve. An average has a real job, because it tells a board which way the business is moving. It cannot tell the people who set the budget where the next dollar should go. That is the question lifetime value exists to answer.
Most companies already track the number. In a 2024 Forrester study paid for by Zeta Global, 81% of the organizations surveyed tracked CLV, yet only 37% used it strategically. The gap is not measurement. It is use.
Two customers join on the same Saturday. One came through a promotion that cut her first-year fee. The other came because her sister had talked about the place for a month. By Monday they share a row on the same report: new member, average value. The report is wrong about at least one of them. Any budget that follows that row will be wrong about her too.
The version of lifetime value sets what a team can decide
Customer lifetime value (CLV) is the value a customer produces over the whole relationship. The simplest version multiplies what she spends per order by how often she buys and by how long she stays. The customer lifetime value guide sets out the formula on margin and a worked calculation.
The definition is correct, and it is also where the trouble starts, because a definition invites a single number and a single number invites a slide. The better test is plain: does the figure change what the company does next for a particular customer? If it does not, it is a report.
Which version of the number a team uses decides which decisions it can make. Revenue CLV, everything a customer has spent so far, flatters the customers who cost the most to serve. It also leaves out the rewards a loyalty program gives away. Margin CLV puts that cost back by subtracting product cost, discounts, rewards, returns, and service, so it shows who has been profitable. Predictive CLV forecasts each customer’s future purchases and margin from her behavior so far. It is the version that can decide anything, because every budget decision is about the future. Finance adds one more step, discounting that future margin to what it is worth today, which is the version that sets acquisition limits.
The forecasting methods are decades old, so the math is the easy part. The harder work is putting the forecast in front of the people who set the budget. Loyalty Juggernaut’s Enterprise Growth Platform starts from the same place: tiers, offers, retention, and breakage should all “bend toward what a customer is worth over time, not what this quarter’s report needs.”
Three public companies report the parts of lifetime value
In annual reports, lifetime value arrives in pieces, one lever at a time. Costco reports what share of its members renew, which is how long they stay. It also reports the share of sales from Executive members, who pay a higher fee for a 2% reward, and that share shows who chooses the upper tier.
Chewy reports how much of its sales comes from customers who use Autoship, its repeat-delivery plan, which is frequency. Its annual report credits high Autoship participation, along with its sales volume and low seasonality, with lowering its cost per unit, which is margin. Ulta Beauty says about 95% of its U.S. sales in fiscal 2025 came from members. Three of the marketing aims it lists read as lifetime value levers.
That last figure, the share of sales a company can tie to a known member, comes before all the others. A lever works one customer at a time, and it works only on customers the company can recognize. Before a team funds any lever, it needs to know how much of its revenue carries a member ID.
Lifetime value decides who gets the next dollar
An average CLV answers the board. The marketing, loyalty, and finance leaders who divide the budget have three different questions, and each one has to be answered customer by customer.
The first is who gets the investment. The customer most likely to respond to an offer can be the one who would have bought anyway. Predicted value shows who will be worth the most. A holdout, a random group of comparable customers kept off the offer, shows whose behavior the offer can still change. Those are two different lists, and a budget built on the first one alone pays for loyalty the company already had.
The second is how much. A renewal incentive should be priced to the value at stake, and the value at stake differs by customer. Tesco reported in April 2026 that every active Clubcard customer’s online grocery journey is personalized one to one. It launched Your Clubcard Prices to 1.5 million customers in March 2026.
The third is when the spending stops. Bain reported in 2001 that Vanguard rejected $40 million from an institutional investor because it expected the money to leave soon, raising costs for every existing customer.
Acquisition belongs under the same rule. Costco’s fiscal 2025 annual report says renewal rates were held down by memberships sold online, including through digital promotions. That is a lifetime value statement about a marketing channel. The channel that brought each Saturday customer in was already a forecast of how long she would stay. A team that keeps the channel on the member record can set a different acquisition limit for each source.
This is also where lifetime value leaves the marketing department. The acquisition budget, the renewal offer, and the reward liability all hang on the same forecast, and so does what an investor will pay for the customer base. When Gupta, Lehmann, and Stuart valued five companies from publicly available data, a 1% gain in retention lifted firm value, what the whole company is worth, by about 5%. That is a finance number, so it belongs in the plan the CFO signs as much as in the deck the CMO presents. Lifetime value is the unit in which loyalty proves it is an enterprise growth engine.
A lever counts when a holdout shows it moved
Member spending grows in two ways: some of it is new, and some of it was always there and now carries a member ID. Identified spending, the purchases tied to a known member, is worth having, since it is what makes lifetime value measurable at all. It is not growth, though, and only new spending raises lifetime value.
A holdout tells the two apart. Withhold the program, or one offer, from a randomly selected group of comparable customers, and measure them against the customers who get it. The gap is the value the program or the offer created, and the rest is value the program recorded.
Three decades of loyalty research add a condition. Programs change behavior more readily than attitude, and how much they change it depends on design and industry. So a program raises lifetime value only when its design gives customers a reason to act differently. The holdout is how a team learns whether it did, before the next budget is set.
Four levers move lifetime value, and referrals add to it
The four levers come from the formula, in the order the customer lifetime value guide uses. Each one moves a metric the business already tracks. Referral value sits beside them, because it is earned through other customers.
- Retention. Keep every customer the business can serve profitably. Retention multiplies the other three levers, because each year she stays is one more year for frequency and order size to do their work. It is also the lever an average hides best, since a program can hold its average while it loses the members who carried it. Track retention by segment and by acquisition channel, and send a win-back offer when a customer’s gap between purchases runs past her usual cycle.
- Frequency. Make the next purchase a habit. A frequency program fits a purchase that already has a rhythm, so time reminders or a subscription to the product’s refill cycle. Dog food runs out on a schedule, which is why a repeat-delivery plan fits a pet retailer. Chewy’s annual report lists high Autoship participation among the reasons its cost per unit is lower.
- Order value. Grow the basket the customer already wants. A tier threshold tells her how the company expects her to shop. Set it a step above how she already buys, and the upgrade feels like recognition. Costco’s Executive tier pays for itself at $3,250 of qualified purchases a year, and its members accounted for 75.6% of fourth-quarter sales. That share shows who chooses the tier, not whether the tier made them spend, so test a new threshold against a holdout before it reaches every member.
- Margin mix. Count the cost of every reward. Rewards, discounts, returns, and service calls all come out of lifetime value. A program that grows revenue per member while shrinking lifetime margin per member is lowering lifetime value, whatever the dashboard says. The reward also sits on the balance sheet as a liability, a debt the company owes members until they redeem, so finance sees its cost before the member does. When a partner funds part of the reward, the customer gets the same value and the margin stays home.
- Referral value. Count who the customer brings. The sister in the opening shows up in the model as one member’s purchases. The customer she brought in does not show up on her record at all, and it should. In a study of bank customers, referred customers were worth more than similar customers who arrived without a referral. Their retention advantage lasted after their margin advantage faded. Count it beside her CLV, and aim referral programs at the segments where referred customers stay longest.
GRAVTY® puts the decision rule to work
In GRAVTY®, each Saturday customer’s record starts with the channel that brought her in. Patented Multi-Dimensional Behavior Tracking then adds her behavior and her influence on others, the sister’s referral included.
Say that three months in, the promotion group visits less often than the referral group. Agentic AI Compass, the team of AI analysts launched in October 2025, reads performance and sentiment data side by side, against benchmarks. It compares the two groups, flags the gap, and tests a renewal offer in simulation before anything ships. GRAVTY’s patented AI-Driven Mass Individualization then shapes the offer for each member. At that scale it is routine: SHARE Rewards, Majid Al Futtaim’s program, issues 4.5 million individualized offers a month on GRAVTY.
The software does not make the decision. The team decides where the budget goes, and every recommendation behind the decision is auditable. The sister who made the referral earns recognition for it. A holdout on the live offer shows which renewals the program created and which it only recorded. The team ends the quarter with a lifetime value decision it can defend, member by member.
The customer never sees the score
She never sees a lifetime value score. She sees a renewal offer that arrived while she was deciding, and a thank-you for the sister who talked her into joining. What she experiences is a company that seems to remember her.
That is the test for every lever above: a tier threshold or a refill reminder should make the relationship more useful to her. A lever that only moves money from one column to another raises the number and leaves the relationship where it was.
The average will keep its place on the board slide, because it describes where the business has been. It never met the two customers who joined on that Saturday, and it cannot tell them apart.
The customer decides her lifetime value one visit at a time. The company decides how many good reasons she has to make the next one.
The numbers behind customer lifetime value
Each company figure below comes from the company’s own filing or results release, with its link.
Four versions of lifetime value, and what each decides
| Version | What it counts | The question it answers | What it decides |
|---|---|---|---|
| Historic revenue CLV | Everything the customer has spent so far | Who has been valuable? | Reporting and recognition |
| Margin CLV | Spend minus product cost, discounts, rewards, returns and service | Who has been profitable? | Reward design and service levels |
| Predictive CLV | Expected future purchases and margin, from behavior models | Who will be valuable? | Where to invest and how much |
| Discounted CLV | Future margin at its present value | What is the customer worth now? | Acquisition limits and valuation |
Three companies and the parts of lifetime value they report
| Company | What it reported | The part of lifetime value it shows |
|---|---|---|
| Costco | Members renewed at 92.3% in the United States and Canada. Membership fees brought in $5.9 billion in fiscal 2026, about half of the company’s $11.7 billion operating income (results, September 24, 2026). | Retention: how long she stays, and what staying is worth |
| Costco | Executive members accounted for 75.6% of its sales in the fourth quarter of fiscal 2026. In the U.S., the tier costs an extra $65 a year for a 2% reward, so the upgrade pays for itself at $3,250 of qualified purchases a year. | Order value: who chooses the upper tier |
| Costco | Its fiscal 2025 annual report notes that renewal rates were held down by memberships sold online, including through digital promotions. | Retention by acquisition channel |
| Chewy | In its September 9, 2026 results, customers who use Autoship, its repeat-delivery plan, accounted for 84.6% of net sales in the quarter. Its net sales per active customer over four quarters came to $602. | Frequency, and a year of buying: frequency and basket together |
| Chewy | Its annual report credits its sales volume, high Autoship participation, and low seasonality with lowering its cost per unit. | Margin: repeat demand lowers the cost to serve |
| Ulta Beauty | Its annual report, filed in March 2026, says about 95% of its U.S. sales in fiscal 2025 came from more than 46 million members. Three of the marketing aims it lists read as lifetime value levers: to “improve guest retention, increase frequency of shopping, and increase spend per member.” | How much of the business it can see customer by customer |
The four levers and referral value, with one tactic each
| Lever | The metric it moves | One tactic |
|---|---|---|
| Retention | Customer retention: the renewal or repeat rate, which sets how long she stays | Send a win-back offer when her gap between purchases runs past her usual cycle |
| Frequency | Purchases per customer per year | Time reminders or a subscription to the product’s refill cycle |
| Order value | Average order value | Set a tier threshold a step above how she already shops |
| Margin mix | Margin per purchase after reward and service costs | Cap reward cost per member, let partners fund part of the rewards, and track the points liability on the balance sheet |
| Referral value | Margin from the customers she brings in, counted beside her own CLV | Aim referral programs at the segments where referred customers stay longest |
Frequently asked questions
How do you calculate customer lifetime value?
Customer lifetime value is average order value times purchases per year, times gross margin, times the years a customer stays. The years she stays equal 1 divided by the yearly churn rate. A shopper who spends $60 a trip, 30 trips a year, at a 25% gross margin and 75% retention is worth $1,800 over four years. The customer lifetime value guide shows each step of the math.
What are the four levers of customer lifetime value?
The four levers of customer lifetime value are retention, frequency, order value, and margin mix. Retention sets how long she stays, frequency how often she buys, order value what she spends each time, and margin mix what is left after rewards and service. Referral value, the margin from the customers she brings in, is counted beside them. A 2007 Harvard Business Review article calls it customer referral value. In a 2004 study of five companies, a 1% gain in retention lifted firm value by about 5%.
What is a good CLV to CAC ratio?
A CLV to CAC ratio of about 3 to 1 is the common rule of thumb. Venture investor David Skok wrote in 2009 that 3x “appears to be a rough minimum for SaaS businesses,” with lifetime value measured on gross margin. He later called such ratios guidelines. Outside software, measure both sides on margin and track how many months it takes to earn back the acquisition cost.
How do you forecast customer lifetime value?
To forecast customer lifetime value, group customers by the month they joined, then predict each one’s purchases from her history. Track spend and retention for each group. Models such as Pareto/NBD, published in 1987, and a simpler version, BG/NBD, published in 2005, use how recently she bought, how often and for how long. Machine learning adds signals such as channel and category. Multiply the predicted purchases by the expected margin, then discount to present value.
Should customer lifetime value be measured on revenue or margin?
Measure customer lifetime value on margin. Revenue CLV favors customers who buy a lot and cost a lot to serve, and it ignores the cost of loyalty rewards. Margin CLV subtracts product cost, discounts, rewards, returns and service, which makes it the version to use for budgets, reward design and acquisition limits.
Does a 5% increase in retention raise profits by 25% to 95%?
No, a 5% gain in retention does not raise profits by 25% to 95% as a general rule. The claim traces to Frederick Reichheld’s work at Bain on service businesses. In his 1990 Harvard Business Review article, cutting defections by 5% raised profits 85% in one bank’s branch system. The gains were 50% in an insurance brokerage and 30% in an auto service chain. A 2001 Bain brief put the effect in financial services at more than 25%. Measure the effect in your own customer data.
Does a loyalty program increase customer lifetime value?
Yes, when the behavior the program changes is worth more than the rewards cost. A 2021 meta-analysis of 429 effect sizes from thirty years of loyalty studies found that loyalty programs reliably change customer behavior. The size of the effect depends on design and industry. The proof for one program is a holdout: withhold it, or an offer, from a random group of comparable customers, and count only the difference as value created.
