Customer Lifetime Value: The Formula, Done Honestly
Nearly every operator I sit down with has an LTV number. Almost nobody can tell me how it was built. It came from a deck, or from somebody multiplying a ticket price by a visit count that felt about right, and then it got pasted into the acquisition budget and never questioned again.
That matters more than it sounds. The customer lifetime value formula is the single number that decides what you can afford to pay for a new client, how hard you should fight to keep an existing one, and what a lapsed client on your list is worth getting back. Get it wrong by a factor of two and every downstream decision is wrong by a factor of two, in a direction you cannot see.
So this is the measurement post. The formula in its three honest forms, two worked examples with every assumption labelled, the mistakes that quietly inflate the number, and how to segment it so it is usable. I am not going to hand you an industry LTV benchmark, because one does not exist in a form worth using. What I will do is make sure the number you compute is yours.
What customer lifetime value is
Customer lifetime value is the total profit a business expects to earn from one customer across the entire length of their relationship with it. It is the sum of every purchase that customer is expected to make, net of the cost of delivering those purchases, over however long they stay active. A med spa client who spends $280 per visit, visits five times a year, stays four years, and comes with a 60% gross margin has a lifetime value of roughly $3,360 in gross profit.
Two words in that definition do all the work. Profit, not revenue. And expects, because LTV is a forecast, not an accounting fact. Every LTV number is a prediction built on assumptions about behavior that has not happened yet, and the quality of the number is entirely the quality of those assumptions.
That is also why LTV sits downstream of retention. Lifespan is the term that moves the answer most, and lifespan is just retention expressed in years. If you have not measured your retention rate, you cannot compute lifetime value, you can only guess at it. How to calculate customer retention rate covers that measurement properly, including the hard part, which is deciding when a silent client counts as gone.
The customer lifetime value formula
There are three versions in circulation. All three are legitimate. They differ in how much they cost you to build and how much they give you back.
Version one: the simple formula
LTV = average order value × purchase frequency × customer lifespan- Average order value (AOV) is average revenue per transaction. In a service business most people call this average ticket or ATV.
- Purchase frequency is the average number of transactions per customer per year.
- Customer lifespan is the average number of years a customer stays active.
Multiply the three and you have revenue per customer over their life. This is the version you will find in most articles about LTV calculation, and it is the one I would not run a business on, for one reason: it tells you what a customer is worth to your top line, and you do not pay for acquisition out of your top line. It is still a fine first pass. A rough number of your own beats a borrowed one.
Version two: the margin-adjusted formula, which is the one to use
LTV = average order value × purchase frequency × customer lifespan × gross marginOne extra term. It changes everything.
Gross margin here means revenue minus the direct cost of delivery: provider labor or commission, product and consumables, card fees, any variable cost that only exists because the appointment happened. Leave out rent, management salary and marketing.
Why it matters: the point of knowing lifetime value is to compare it against what you spend to acquire or recover a customer, and acquisition spend comes out of gross profit. Comparing a revenue-based LTV to a cash acquisition cost compares two numbers denominated in different things. In a service business running a 55% to 65% gross margin, the revenue version overstates a customer by roughly 40%, which is enough to make a bad channel look profitable. If you only build one LTV number, build this one.
Version three: the discounted and predictive versions
Money that arrives in year four is worth less than money that arrives this month, so the rigorous version discounts future contribution back to present value:
LTV = Σ over each year t of (annual gross profit per customer in year t ÷ (1 + d)^t)where d is your annual discount rate. The common shortcut you will see for a steady-state subscription business is:
LTV = (ARPA × gross margin) ÷ (churn rate + discount rate)Beyond that sit the genuinely predictive models: cohort-based survival curves, and probabilistic approaches like BG/NBD paired with a gamma-gamma spend model, which estimate each customer’s future transactions from their own purchase history rather than from a blended average.
Worth it if you are a subscription or ecommerce business with tens of thousands of customers, clean transaction-level data, a lifespan measured in many years, and someone in-house to maintain the model. Also worth it if you are raising money against your LTV, since investors will discount it anyway.
Not worth it for most multi-location service operators, and I will say that plainly because the LTV software category exists mostly to sell you otherwise. If your average client lifespan is two to four years, discounting shifts the number modestly, well inside the error bars on your lifespan estimate. You are applying three decimals of precision to an input you measured to one. Measure lifespan properly instead. The exception is a long payback decision, say acquisition that only breaks even in month twenty, where the whole question is the time value of that money.
Worked example one: a multi-location service business
Illustrative numbers throughout. These are not benchmarks, they are inputs chosen to show the mechanics. Substitute your own.
Assumed inputs, a three-location med spa, injectables and facials blended:
| Input | Assumed value | Where it comes from |
|---|---|---|
| Average ticket (AOV) | $240 | Total service revenue ÷ transactions, last 12 months |
| Visits per client per year | 3.4 | Transactions ÷ distinct active clients |
| Average client lifespan | 2.8 years | Derived from retention, see below |
| Gross margin | 58% | Revenue less provider comp, product, card fees |
Simple version:
$240 × 3.4 × 2.8 = $2,284.80 in lifetime revenueMargin-adjusted version:
$240 × 3.4 × 2.8 × 0.58 = $1,325.18 in lifetime gross profitCall it $1,325. That is the number to use, and note it is 42% smaller than the one most operators quote. If this business were paying $400 to acquire a client and comparing it against $2,285, it would believe it was running at better than 5x. It is actually running at 3.3x. Still fine. But if acquisition cost crept to $700, the revenue-based view would still look healthy at 3.3x while the real ratio had fallen under 2x, and nobody would notice until cash got tight.
Where lifespan came from. This is the input people invent, so here is the arithmetic. At 64% annual retention:
Average lifespan ≈ 1 ÷ (1 - 0.64) = 2.78 yearsThat 1 ÷ churn shortcut assumes constant churn every year, which is never true. Real churn is front-loaded, heaviest in the first few visits. Treat it as a starting estimate and replace it with cohort data once you have eighteen months: take the clients who first visited in a given month and track what share are still active at twelve, twenty-four and thirty-six months. That curve is your lifespan, and it will be shorter than the shortcut suggests.
The segment the blended number hides. In the client files we audit, a large share of service clients never return after their first or second visit. Average them in with established regulars and you get a mid-range figure that describes nobody. Run it split:
| Segment | Assumed lifespan | Margin-adjusted LTV |
|---|---|---|
| First-visit-only clients | 0.1 years, one visit | $139 |
| Clients who reach visit three | 4.1 years | $1,940 |
Same business, same formula, two numbers an order of magnitude apart. The decision the blended figure cannot make: how much is converting a first-timer into a third visit worth? On these illustrative numbers, about $1,800 of gross profit per client. That reframes a phone call from a cost into an obvious trade.
Worked example two: subscription and ecommerce
The formula is the same. The inputs come from different places and the churn term does more of the work.
Assumed inputs, a replenishment ecommerce brand with a subscribe-and-save option:
| Input | Assumed value |
|---|---|
| Average order value | $62 |
| Orders per subscriber per year | 8 |
| Monthly logo churn | 5.5% |
| Gross margin after COGS, fulfilment, payment fees | 46% |
Annual retention from 5.5% monthly churn:
(1 - 0.055)^12 = 0.509, so 50.9% annual retention
Lifespan ≈ 1 ÷ (1 - 0.509) = 2.04 yearsMargin-adjusted LTV:
$62 × 8 × 2.04 × 0.46 = $465.32Two things to flag, because both are where ecommerce LTV numbers go wrong.
Discounts belong in the margin line, not in a footnote. If 40% of those orders go out on a 20% subscriber discount, your effective AOV is not $62 and your margin is not 46%. Compute both net of discount actually given, or you build an LTV on list prices nobody paid.
Failed payments are churn. A subscriber whose card declines and who never updates it has churned, whatever the dashboard says, and involuntary churn is often a large slice of the total. Dunning management covers recovering those, and subscription churn covers separating voluntary from involuntary. On the retention side, ecommerce customer retention is the companion to this post.
For a subscription model, the steady-state shortcut gets you close:
($62 × 8 × 0.46) ÷ (0.491 + 0.10) = $386using 49.1% annual churn and a 10% discount rate. Lower than the undiscounted $465, as it should be, and that gap is roughly the cost of waiting for the money.
Where operators get LTV wrong
Five mistakes. I see all five regularly, and each one pushes the number in the same direction, which is up.
1. Using revenue instead of margin. The most common and the most expensive, because it inflates LTV by exactly your cost of delivery, then gets compared against a cash CAC. Fix: multiply by gross margin. One cell.
2. Assuming a lifespan nobody measured. Somebody picks three years because it sounds reasonable, and that input is doing more work in the formula than any other. Derive it from retention, then replace the derivation with cohort survival data. If you must guess, guess low and revise up.
3. Averaging across segments that behave nothing alike. A company-wide LTV blends first-timers with ten-year regulars, members with walk-ins, your best location with your worst. If your segment-level LTVs vary by more than roughly 2x, the blended figure is not a summary, it is noise.
4. Letting outliers set the average. One client who bought a $9,000 treatment package drags the mean up and makes a channel look viable when it is not. Check the median against the mean. If they are far apart, report the median, and look at the distribution before trusting any single figure.
5. Double counting a reactivated customer. Specific to anyone running win-back. A client lapses, you recover her eleven months later, and your system creates a second client record or your cohort logic treats her as new. Her spend now sits in two lifetimes, both look healthy, and true per-relationship LTV is overstated. Dedupe on phone number, keep the original first-visit date, and record reactivation as an event inside one continuous relationship rather than the start of another.
A sixth worth a line: counting revenue invoiced but not delivered. A prepaid series of six treatments is one transaction and six future obligations. Recognize it across the visits, not at purchase, or your frequency and margin will both be wrong.
Segmenting LTV so it is actually actionable
A company-wide LTV is a board slide. A segmented one is an operating tool. Four cuts, roughly in order of what they usually reveal.
By location. Always first in a multi-unit business. Two locations at $1,600 and one at $700 average out to something acceptable-looking, and the $700 site is where your acquisition dollars quietly die. Rank the table. Put names on it.
By service line. Membership and prepaid-series clients should show structurally higher LTV than single-visit clients, because the decision is made in advance and the next visit is already paid for. Quantifying that gap turns “we should push memberships” into a specific dollar value per enrolment for the checkout conversation.
By acquisition channel. The cut that most often changes a budget. Deep-discount introductory offers produce clients with worse frequency and shorter lifespans than full-price ones. If you only track cost per first visit, the discount channel wins forever, because the damage shows up in a lifespan nobody measures. Compute LTV and CAC per channel and some channels stop looking like growth.
By first-visit cohort. Group clients by the month of their first visit and track value accumulation per cohort. This separates “our LTV is falling” from “the clients we acquired during last spring’s promo were worse, and everything since has been fine.” Period-level averages cannot make that distinction.
LTV to CAC and payback period
Once LTV is in gross-profit terms, two derived numbers do the real work.
LTV:CAC ratio = margin-adjusted LTV ÷ fully loaded customer acquisition cost
CAC payback period = CAC ÷ monthly gross profit per customerFully loaded CAC means all of it: media spend, agency fees, the salary share of whoever runs the campaigns, the introductory discount you gave away. Dividing media spend alone by new clients is how a 2:1 business reports 5:1.
On the ratio itself: 3:1 is widely cited as a healthy target across SaaS and agency writing. It is a rule of thumb, not a research finding, and as far as I can tell no empirical study sits behind it. Treat it as a sanity check with a direction, not a target with authority. Under about 1:1 you are paying more than a customer is worth and the only question is how long you can fund it. Comfortably above 3:1 more often signals underinvestment in growth than excellence, because if every customer is that profitable you could probably buy more of them.
Payback period is the number I watch more closely than the ratio, because it governs cash. A business with great LTV and an eighteen-month payback can run out of money while being fundamentally sound. If payback runs longer than a few months in a service business, the LTV argument behind your acquisition spend needs to be much more solid, and that is exactly where the discounted formula earns its keep.
How to increase customer lifetime value
Three levers, and only three. Raise what a client spends per visit, raise how often they come, or extend how long they stay. Everything in a retention deck resolves to one of those. The useful question is which ones actually move in a service business.
Lever one: raise average order value
Realistically movable, in bounded amounts. Service-line attachment is the honest version: a complementary service or a retail product at checkout. It needs provider training and a script, but it works and it compounds across every future visit.
Price increases also raise AOV and are the fastest route to a higher LTV on paper. They also affect frequency and lifespan in ways that take two quarters to appear. If you raise prices, watch visit frequency per client for six months before you book the LTV gain.
Mostly wishful: a 30% AOV lift from premium packaging or a tier rename. Perceived value shifts AOV at the margin, not in step changes.
Lever two: raise purchase frequency
The most movable lever in a service business, and the most underworked. Frequency is largely a function of whether the next appointment exists. Rebooking at checkout, before the client leaves the building, does more work than anything else here. The gap between locations that do it as policy and locations that do it when they remember shows up in the frequency column.
Recall workflows for clients who did not rebook are the backstop. Membership and prepaid structures are stronger still, because they remove the decision entirely: the visit is already bought. That is the real case for a membership or loyalty structure, and it is an LTV argument, not a marketing one.
Mostly wishful: expecting a client on a natural eight-week cycle to come every five weeks because you emailed more. Service cadence has a physical floor. Email frequency does not change hair growth.
Lever three: extend lifespan
The highest-value lever and the hardest. It is also where most of the compounding sits, since lifespan is multiplicative in the formula.
What moves it, in my experience: surviving the first two visits, where most of the loss happens; provider continuity, because the relationship belongs to a person and a stylist’s departure is a churn event; and catching silence early, because a client four weeks past their expected rebooking date is recoverable and the same client at six months mostly is not.
Mostly wishful: a win-back email to a client who lapsed eleven months ago. Not zero, but close enough to zero that it should not be in your plan as the main mechanism. The retention strategy hub organizes the full set of levers by lifecycle stage, and reducing customer churn covers the prevention side. If you want the vocabulary underneath all of it straight, what customer churn is is the definitions post. For the patient version of lifespan in a dental or DSO context, patient retention goes deeper, and referral program ideas covers the one adjacent lever that raises LTV indirectly by lowering CAC on the clients a good relationship brings in.
One honest ranking for an operator starting from nothing: frequency first, because rebooking at checkout is free and immediate. Lifespan second, through first-visit follow-up and lapse detection. AOV third, because it needs training to stick and it is the one most likely to be given back in discounts.
Tooling: do this in a spreadsheet first
There is a customer lifetime value software category and some of it is good. You almost certainly do not need it yet.
Everything in this post runs off one export from your booking platform or CRM: client ID, location, service line, first-visit date, last-visit date, and transaction-level amounts. Mindbody, Zenoti, Vagaro, Boulevard and the dental PMS systems can all produce that. From those columns you can compute AOV, frequency, cohort survival, lifespan, segmented LTV and LTV:CAC in one spreadsheet in an afternoon.
Some of those platforms ship a built-in LTV widget. Before you trust one, find out exactly what it computes. In my experience it is usually cumulative historical revenue per client, which is lifetime-to-date revenue: undiscounted, pre-margin, backward-looking. A useful column, not this metric. If the documentation will not state the definition, assume that is what it is and compute your own. Customer retention software covers the platform landscape.
Buy an LTV or CDP tool when three things are true: transaction volume past the point where a spreadsheet rebuild hurts, channel-level budget decisions made weekly rather than quarterly, and someone who owns the number. Until then the tool is a way to avoid agreeing on the definition, and the definition is where the value is.
If you want the reactivation-specific slice of this math run against your own list, the ROI calculator does it from your list size, ATV, and lapse window.
How LTV sets the ceiling on a reactivated customer
Margin-adjusted LTV tells you the maximum you can rationally spend to start or restart a customer relationship. For acquisition, everyone understands that. For reactivation, almost nobody applies it, which is why lapsed lists sit untouched in CRMs while the same operators bid aggressively on cold traffic.
On the illustrative figures above, a client who reaches visit three carries roughly $1,940 of gross profit. Recovering a lapsed client who already knows your brand needs no awareness spend, no introductory discount, and no onboarding, and recovered clients in our campaign data rebook at full price far more often than new ones do. The LTV you just calculated is the ceiling on what that recovery is worth, and it is a high ceiling relative to what reactivation costs.
That is the entire economic argument, and I am not going to re-run it here because customer reactivation vs new customer acquisition ROI works the comparison properly with both cost structures side by side. What this post adds is the denominator. That comparison is only as good as the LTV you put into it, which is why doing the margin adjustment and the segment split matters before you use it to move a budget.
Two things to be careful about when you apply LTV to a lapsed list.
A reactivated client’s remaining lifetime is not a full lifetime. She has already consumed part of her relationship, so what you are buying is the remainder. Model it as remaining value, and measure post-reactivation lifespan separately where you have the data. In our campaigns recovered clients tend to behave like established clients rather than new ones once they are back, which is favorable, but measure it rather than assuming it.
The attribution window has to be set in advance. If a recovered client books three weeks after the call, is that the campaign? Six weeks? Decide before the campaign runs, not after you see results. We price on recovered revenue, so we agree that window with every client up front, and I would hold any reactivation vendor to the same standard. The reactivation guide covers how the rate and the window are constructed, and our churn statistics page collects what we can source, including our own campaign data, with the sourcing stated.
Customer lifetime value FAQ
What is the customer lifetime value formula?
The practical formula is: LTV = average order value × purchase frequency per year × customer lifespan in years × gross margin. The first three terms give lifetime revenue per customer, and multiplying by gross margin converts it to lifetime gross profit, which is the form you need in order to compare it against acquisition cost. The simple version omits gross margin and overstates what a customer is worth by the size of your delivery cost.
How do you calculate LTV?
Pull four inputs from your booking or ecommerce data. Average order value is total revenue divided by transactions. Purchase frequency is transactions divided by distinct active customers, over a year. Customer lifespan is derived from your annual retention rate, approximately 1 ÷ (1 - retention), replaced with cohort survival data once you have eighteen months of history. Gross margin is revenue less only the variable cost of delivery. Multiply all four. Then compute it again split by segment, because a blended figure across first-timers and established regulars describes neither.
What is a good LTV to CAC ratio?
The widely cited rule of thumb is 3:1, meaning a customer should generate three times the fully loaded cost of acquiring them. Treat it as a heuristic rather than a researched benchmark, because it has no real study behind it. Below roughly 1:1 you are paying more than a customer is worth. Far above 3:1 usually means you could profitably acquire more customers than you are. Watch CAC payback period alongside it, since a healthy ratio with an eighteen-month payback still creates a cash problem.
What is the difference between LTV and lifetime revenue?
Lifetime revenue is total gross spend from a customer. LTV, used properly, is the gross profit from that spend, and ideally discounted to present value. The distinction matters because acquisition spend is funded from gross profit, so comparing lifetime revenue to acquisition cost flatters every channel. Many booking platforms label cumulative historical revenue per client as LTV. That is lifetime-to-date revenue, which is a useful column but not this metric.
How do you increase customer lifetime value?
Only three levers exist: raise average order value, raise purchase frequency, or extend customer lifespan. In a multi-location service business frequency is usually the most movable, mainly through rebooking at checkout and recall workflows before a client drifts. Lifespan carries the most compounding value and moves through surviving the first two visits, provider continuity, and catching lapse early. AOV moves in bounded amounts through service attachment and retail. Expecting email frequency to compress a natural service cadence is the most common wasted effort.
The short version
Customer lifetime value is AOV × frequency × lifespan × gross margin. The margin term is not optional, and lifespan is the input that deserves the measurement work because it is the one the answer is most sensitive to.
Compute it segmented, or do not bother. By location, by service line, by acquisition channel, by first-visit cohort. If your segment LTVs differ by more than about 2x, the blended number is noise with a dollar sign on it.
Then use it for what it is for: setting a ceiling on what you can spend to start or restart a relationship. Export client ID, first-visit date, last-visit date and transaction amounts. Build the four inputs. Split first-timers from established clients and look at the gap. The clients that export says you lost are in the same file with phone numbers attached, and the LTV you just calculated is what each one is worth getting back.