# Customer Lifetime Value Customer lifetime value is the margin a customer is expected to produce over the relationship, used to set what winning and keeping them is worth. **The question.** What is a customer of this type worth over the relationship, and so how much can we spend to win or keep them? Also known as: CLV. Also written LTV or CLTV. Customer equity is the sum of CLV across the base. ## When to use it - Acquisition spend is set from last year's budget, or from cost per lead, with no link to how long a customer stays or what they contribute. - Two channels or two segments show similar close rates and very different retention, and both are funded as if they were worth the same. - A loyalty or save offer is being priced, and nobody can say how much retained margin is at stake. - A subscription business is growing revenue and shrinking the quality of the base, and you need a stop rule for paid acquisition. ## When to avoid it - There is no usable retention or margin history, as with a first launch. A test of whether anyone pays is the work, not a lifetime model. - Relationships are one-off, as with many project businesses. Expected periods are one, and contribution on the job is enough. - Customers are not free to leave, as with a regulated monopoly. Retention is not a choice the spend can change. - Someone wants a single firm-wide number to put in a board pack. CLV belongs to a segment or a cohort. An average of those hides the decision. ## Overview Customer lifetime value is the contribution a customer is expected to produce over the rest of the relationship, discounted. The simple form is margin times retention, divided by one plus the discount rate minus retention. An even simpler form multiplies contribution by expected remaining periods and skips the discount. The choice it supports is how much to spend to win a customer, or to keep one, by segment. The number belongs to a cohort, not to the firm. A blended CLV funds the leaky channel with the sticky channel's economics. Use contribution, not revenue, keep acquisition cost outside the remaining-value figure, and treat constant retention as an assumption to test, not as a fact about how people stay. ## The inputs ### Margin per period Contribution from the customer in one period, after variable cost of serving them, not revenue and not fully loaded profit. Acquisition cost sits outside the remaining-value formula and is compared with it. Using revenue as the "margin" overstates what winning a customer is worth. ### Retention rate The share of customers who are still there at the end of the period. In a contractual book it is 1 minus churn. A rate that is an average of a sticky group and a leaking one will price both groups wrong. ### Discount rate The cost of waiting for later margin, usually the firm's cost of capital or a hurdle close to it. It stops a long tail of hoped-for renewals from looking as good as cash this year. ### Expected remaining periods The undiscounted cousin of the retention factor. If a customer stays with probability r each period, expected remaining periods after this one are r / (1 - r). Contribution times that count is the no-discount form. It is easier to explain and too high once money in year five is treated as equal to money now. ## What it is not - Revenue per customer times years on the books. That ignores cost to serve, the chance they leave, and the time value of later years. - A number that already nets off acquisition cost. Remaining CLV is what they will produce. Subtract CAC to see whether winning them pays. Mixing the two hides both decisions. ## The steps 1. **Segment before you average.** Compute CLV by the groups that actually differ, such as channel, offer, or customer type. A single firm-wide figure will fund the worst group with the best group's economics. 2. **Get margin and retention from the same cohorts.** Use contribution after cost to serve, and retention from the same periods and groups. Do not mix a new-logo retention rate with the margin of long-stayers. State the period length, usually a year for subscriptions and a quarter for higher churn. 3. **Compute remaining CLV and compare with CAC.** Apply the simple formula, and show the undiscounted version beside it so people can see what discounting did. Subtract nothing yet. Then put CAC next to remaining CLV for each group. 4. **Test the two or three inputs that move it.** Re-run at retention a few points up and down, and at a discount rate a point off. If the accept-or-stop decision flips inside that range, say so. Do not report CLV to the nearest dollar. 5. **Set a spend rule and a review date.** Cap acquisition where CAC exceeds remaining CLV, and price save offers against the CLV at risk. Refresh cohort retention on a fixed calendar, because the number is a snapshot of how that group has been behaving. ## The formula CLV = m × r / (1 + d - r), or CLV = contribution × expected remaining periods Remaining lifetime value is this period's margin, times retention, divided by the discount rate plus the chance they leave. The second form skips discounting and multiplies contribution by how many periods they are expected to stay. - m: contribution margin per customer per period, after variable cost to serve, before acquisition cost. - r: retention rate per period, as a decimal, such as 0.75 for 75 percent. Must be below 1 + d. - d: discount rate per period, as a decimal, such as 0.10 for 10 percent. - Expected remaining periods: r / (1 - r) after the current period, or 1 / (1 - r) if the current period is already certain. Use this only in the undiscounted form. - Acquisition test: spend up to remaining CLV, not above it, unless you have a separate reason to buy the customer at a loss. ## In practice A B2B software firm spends $1,200 to acquire a customer through paid search. Gross margin is $800 a year. Annual retention is 75 percent. Finance's discount rate is 10 percent. Remaining CLV = 800 × 0.75 / (1.10 - 0.75) = 600 / 0.35 = $1,714. The $1,200 CAC is under that, so the channel looks fine on the average customer. The undiscounted cousin, contribution times expected remaining years, is 800 × (0.75 / 0.25) = $2,400, which is the figure the marketing deck has been using. Discounting is what closed most of the $686 gap. Split the book. Customers from search retain at 60 percent. CLV = 800 × 0.60 / (1.10 - 0.60) = 480 / 0.50 = $960, which is below the $1,200 CAC. Customers from partner referrals retain at 85 percent. CLV = 800 × 0.85 / (1.10 - 0.85) = 680 / 0.25 = $2,720. Search is buying customers worth less than they cost. The recommendation is a cap on search at the 60 percent cohort, not a higher blended CAC target, and a shift of the same budget toward partners. A 5-point retention lift on the search cohort, 60 to 65 percent, only raises CLV to 800 × 0.65 / 0.45 = $1,156, still short of $1,200, so a save program there has to beat that math or the spend should stop. ## What you leave with - Remaining CLV by segment or cohort, with m, r, and d shown. - The undiscounted contribution × periods figure beside it, so discounting is visible. - CAC by the same groups, and a cap or stop where CAC exceeds CLV. - A sensitivity on retention, which is usually the input that flips the decision. - A review date for the cohort rates. ## Failure points ### Revenue standing in for margin Top-line spend per customer is treated as value. Cost to serve is ignored, so high-touch accounts look like the ones to buy more of, and they destroy contribution. ### One average for the base A blended 75 percent retention funds a 60 percent channel with a 85 percent story. The average is true and the spend decision is wrong. ### CAC folded into CLV The model subtracts acquisition cost inside the lifetime figure, then someone compares that net number with CAC again, or uses it as a save-offer ceiling. Remaining value and acquisition cost have to stay separate. ### Retention from survivors The rate is taken from customers still on the books, which drops those who already left. r is too high, CLV is too high, and paid acquisition expands into a leak. ### A tail that never ends Infinite-horizon math is applied to a product with a three-year life, or to a cohort whose retention is already falling. Truncate, or use a declining retention path, when the constant-r assumption is the fiction. The failures above are usually [Commercial Acumen](https://usefulconsultingframeworks.com/your-edge/#commercial): the work was not tied to money, for the client and/or for you. ## Origin Valuing a customer as the profit expected over the relationship is older than any one paper. Catalog, insurance, and book-club businesses were already asking what a new name was worth, against what it cost to get one, well before the term settled. F. Robert Dwyer's 1989 Journal of Direct Marketing article, "Customer Lifetime Valuation to Support Marketing Decision Making," is an early formal treatment in that literature. Robert Blattberg and John Deighton brought the idea to general managers in "Manage Marketing by the Customer Equity Test," Harvard Business Review, July 1996, as a way to split spend between winning customers and keeping them. Paul Berger and Nada Nasr set out a family of simple closed-form models in "Customer Lifetime Value: Marketing Models and Applications," Journal of Interactive Marketing, Winter 1998, including the constant-margin, constant-retention, infinite-horizon form used below. Later work by Sunil Gupta, Donald Lehmann, and others, and by Peter Fader and Bruce Hardie, added migration, always-a-share, and buy-till-you-die models for non-contractual settings. There is no trademark on the idea. The formula you use is a choice about how customers actually leave. ## Further reading - [Customer lifetime value](https://en.wikipedia.org/wiki/Customer_lifetime_value). Wikipedia's overview of the simple formula, the discounting, and the gap between contractual and non-contractual models. Free. - [Manage Marketing by the Customer Equity Test](https://hbr.org/1996/07/manage-marketing-by-the-customer-equity-test). Robert Blattberg and John Deighton, Harvard Business Review, July 1996. The short managerial argument for splitting spend between winning customers and keeping them. Paywalled. ## Related frameworks - [Net Promoter Score](https://usefulconsultingframeworks.com/frameworks/nps/): A loyalty proxy from a survey. Pick it to find who is unhappy and why. Pick CLV to put a dollar ceiling on what winning or saving a customer is worth. - [Price Elasticity of Demand](https://usefulconsultingframeworks.com/frameworks/price-elasticity/): Sizes volume lost or gained from a price move. A higher price can lift margin per period and cut retention. Run both before treating a price rise as a CLV win. - [Marketing Funnel](https://usefulconsultingframeworks.com/frameworks/marketing-funnel/): Shows where prospects and customers leave. CLV sets what a retained customer is worth. The funnel shows which stage the leak is in. - [Brand Equity](https://usefulconsultingframeworks.com/frameworks/brand-equity/): Explains part of why some groups stay and pay. Pick it for the source of stickiness, and CLV to size it. - [Break-Even and Contribution Margin](https://usefulconsultingframeworks.com/frameworks/break-even/): Finds the volume at which contribution covers fixed cost. Pick it for a product or a plant. Pick CLV for a customer relationship. Last updated October 20, 2026.