Contents
In short. LTV (customer lifetime value) is the money a customer brings to a business over a chosen period. For advertising decisions, calculate it from margin (after returns and variable costs) and from a cohort you observed, not from average order value times average order count. In the illustrative example below, a cohort's 12-month LTV is $100.80 by revenue and $45.36 by margin, against a $40 acquisition cost. The "LTV three times CAC" rule is, in the primary sources we could open, a subscription guideline with no method behind it.
What LTV is and why it has three different values
LTV is the money a customer brings over the whole relationship. Shopify calls its basic formula revenue-based and adds a margin-adjusted version. In academic work it is the present value of the future cash flows attributed to the customer relationship, as Fader and Hardie quote Pfeifer and co-authors (2005).
One number hides three decisions: what to count (revenue or margin), over what period, and for which group. There are three methods (by revenue, by margin, by observed cohort), plus a churn formula for subscriptions.
Scroll right to see all columns
| Method | Formula | Honest when | What it hides |
|---|---|---|---|
| By revenue | Order value x orders x lifespan | A quick estimate | Returns and costs: overstates what is left |
| By margin | The same x margin share | Pricing a customer or lead | Timing of the money |
| Subscription, by churn | Revenue per account x margin / churn | Stable churn, visible cancellations | Constant churn is assumed |
| By cohort | Group margin over N months / customers | You have order history | Only a window, not a lifetime |
Why average order value times average order count misleads
The average of 1.68 orders matches nobody
The formula "order value x orders x lifespan" is valid arithmetic when all three numbers come from one group over one period: a $60 order and 1.68 orders give $100.80 per customer. The danger is that an average describes a group, not a customer. Fader and Hardie warn that averages make us ignore variability, which leads to bad decisions.
Illustrative example. A store acquired 250 customers in March and watched them for 12 months; every order is $60. 160 customers (64%) bought once, 50 twice, 25 three times, 10 five times and 5 seven times. The average of 1.68 orders matches nobody. The 40 customers with three or more orders, 16% of the cohort, brought 38% of its margin.
The second trap is mixing groups: order value from all orders, order count from loyal customers, lifespan from the oldest. The result describes nobody you acquired. Shopify warns that a healthy blended LTV-to-CAC ratio can hide a channel that attracts low-value customers at a high cost.
Why LTV without margin and time period misleads
Revenue is not the business's money. Margin (contribution in management accounting) is what is left after variable costs; per OpenStax, it covers fixed costs first and the rest is profit. In the example, 45% of each $60 order is left after returns, product and delivery: $27.
The whole cohort over 12 months: LTV is $100.80 by revenue and $45.36 by margin. CAC, the full cost of acquiring a customer, is $40 ($10,000 for 250 people). The LTV-to-CAC ratio is 2.52 by revenue and 1.13 by margin, because 2.52 x 0.45 = 1.13. The first looks like a miss against 3:1; the second says the cohort has paid back, with a thin cushion.
Time is the other half. Right after the first purchase cohort LTV is $27, after three months $33.48, after a year $45.36. Fader and Hardie note that a customer's true value is unknown until the customer has gone, and a calculation cut off at a horizon is better called truncated.
How to calculate LTV by cohort: an illustrative table
A cohort is a group of customers with a common trait, here the month of first purchase. Cohort LTV is an observation, not a forecast: take everyone who first bought in March and add up their margin per customer. Shopify advises comparing cohorts over equal periods, since older groups had longer to buy. Group age is covered in "Why lead-to-sale conversion is 3% in one report and 12% in another".
Conditions: 250 customers, $60 per order, 45% margin ($27 per order), acquisition cost $10,000, CAC $40. The numbers are invented, not a company result or a norm.
The cohort passes CAC during the 8th month if purchases in months 7-9 are spread evenly (an assumption). Over 12 months it earns $11,340 of margin on $10,000 of acquisition cost, leaving $1,340. A four-month-old cohort cannot be compared with this one.
| Item | Cumulative margin per customer | CAC |
|---|---|---|
| First purchase | $27 | $40 |
| Mo. 1-3 | $33.5 | $40 |
| Mo. 4-6 | $38.3 | $40 |
| Mo. 7-9 | $42.1 | $40 |
| Mo. 10-12 | $45.4 | $40 |
Illustrative example: the cohort passes CAC between months 7 and 9
Scroll right to see all columns
| Period | Orders | Revenue, $ | Margin, $ | Cumulative per customer, $ | After CAC, $ |
|---|---|---|---|---|---|
| First purchase | 250 | 15,000 | 6,750 | 27.00 | minus 13.00 |
| Months 1-3 | 60 | 3,600 | 1,620 | 33.48 | minus 6.52 |
| Months 4-6 | 45 | 2,700 | 1,215 | 38.34 | minus 1.66 |
| Months 7-9 | 35 | 2,100 | 945 | 42.12 | 2.12 |
| Months 10-12 | 30 | 1,800 | 810 | 45.36 | 5.36 |
How LTV turns into an affordable cost per customer and per lead
LTV by margin shows how much you can spend on a customer once you decide what share of margin the business keeps. At 25% the affordable CAC is $45.36 x 0.75 = $34.02, and the actual $40 is $5.98 above it. Acquisition cost is not subtracted from LTV when estimating an upper bound on acquisition spending (Fader and Hardie).
As in "How much can you pay for a lead and a customer and still stay profitable", subtract non-advertising cost and multiply by the share of leads that buy. With $6 of other costs and 8% of leads buying, the ceiling per lead is ($34.02 - $6) x 8% = $2.24, under the example's conditions, not a forecast. CPL, CAC and ROAS are covered in "A cheap lead can still be an expensive customer".
Value in an ad platform is your decision: Google Ads lets you set conversion values such as sales revenue or profit margins, and you enter them yourself. For new customers it has customer lifecycle goals: New customer value uses value-based bidding and needs a purchase goal; New customer only also fits lead generation. The platform does not know your margin or payback period: carry them into each channel's affordable CAC.
Where the 3-to-1 rule comes from and whether you can rely on it
Derive your own threshold from margin, cash and time
In the primary sources we could open, the rule is a subscription guideline with no method behind it, not a norm. The earliest mention we could open is David Skok's "Startup Killer: the Cost of Customer Acquisition" (no date on the page): a very rough rule of thumb that LTV should be about three times CAC for a subscription model, closer to five for public companies, with no calculation or data.
Secondary blogs give different dates and authors without a method, so we do not claim an origin beyond Skok's guideline.
The guideline carries a caveat: in Skok's definitions it assumes the simpler LTV formula without a margin adjustment and a gross margin of 80% or higher. With a 45% margin, "three by revenue" equals 1.35 by margin. A ratio without a payback period also says little: 3 to 1 can be repaid in half a year or in five years. Derive your own threshold from margin, cash and time.
Scroll right to see all columns
| Source | What it says | What is missing |
|---|---|---|
| Skok, Startup Killer (no date) | LTV about 3x CAC for a subscription model; a very rough rule of thumb | Calculation, data |
| Skok, SaaS Metrics 2.0 (update note 05/2016) | Best SaaS companies above 3, sometimes 7 or 8; many healthy ones miss it early | Method, sample |
| Skok, article on discounting | Earlier advised above 3; discounted LTV is lower and he promised to update | A new value |
| Shopify, CLV blog (Oct 8, 2026) | 3:1 is the benchmark, the target depends on margins and costs | Method, industry, study date |
Where an LTV calculation stops working
LTV is an estimate, not a guarantee. State the limits before a budget decision.
- Little data. If the 5 customers with seven orders had bought once, cohort LTV would be $42.12 instead of $45.36. Look at several cohorts.
- Subscriptions and one-time purchases. Churn formulas work when leaving is visible; in a store customers just stop buying, and a repeat-purchase rate is not retention (Fader and Hardie). Churn within a cohort is not constant, and lifetime as 1 / churn holds only if it is.
- Returns. A return arrives after the sale, so a fresh cohort looks better than it will end up. Set a return window.
- Tools differ. ChartMogul computes LTV as revenue per account over churn, without margin; the former Google Analytics report capped "lifetime" at 90 days and was deprecated on 29 January 2024.
- Discounting. Future money is worth less; agree the rate with finance (Fader and Hardie). The table has no discount.
Assignment for the team: LTV by cohort in one working day
The result: a table of cohorts by first-purchase month with margin per customer, CAC and payback period per channel. Until it exists, the budget does not grow on a blended LTV.
Data: Google Analytics 4 cohort exploration can group users by first touch or first transaction, shows at most 60 cohorts and uses device data only, so it does not replace order history with returns and costs.
- Define customer, first purchase, observation period (say, 12 months) and return window.
- Export orders: customer, date, amount, refund, variable costs, source of the first purchase.
- Group customers by month of first purchase, each in one cohort.
- Compute cumulative margin per customer at 3, 6, 9 and 12 months; compare only cohorts of one age.
- Add the cohort's CAC and find when cumulative margin caught up.
- Split by channel; find the share of one-time buyers and each channel's affordable CAC.
- Record the decision (raise, hold, change the offer) and the recalculation date.
How to check an LTV number a contractor shows you
Ask for the same table: cohort and its age, revenue or margin, make-up of CAC, returns. Agency forecasts are covered in "An agency sent a lead forecast: how to tell whether to trust it".
Disclosure: the journal is published by a project connected with the marketing company WEBNOSTRA; the same questions apply to it.
Key decisions
- calculate LTV from margin and from a cohort of one age; keep revenue as an upper bound
- compare LTV with CAC together with payback period, channel by channel
- treat 3-to-1 as a subscription guideline and derive your own threshold from margin, cash and time
What to discuss with your team
- It is stated whether LTV is by revenue or margin and what was subtracted
- The number has an observation period, and only cohorts of one age are compared
- LTV is calculated per channel cohort, not as one store-wide average
- Acquisition cost is not subtracted from LTV twice, and CAC matches other reports
- A comparison with 3-to-1 names the basis (margin or revenue) and the period
- A recalculation date and a no-growth threshold are set
Terms in plain language
- LTV (customer lifetime value)
- The money a customer brings over a chosen period, by revenue or by margin
- CAC
- The full cost of acquiring one new customer, with a named cost scope
- Cohort
- Customers with a common trait, such as the month of first purchase
- Margin (contribution)
- Selling price minus variable costs; it covers fixed costs first
- Churn
- The share of customers who left in a period, where leaving is visible
- Payback period
- The time in which cumulative margin catches up with CAC
Sources
- David Skok, forEntrepreneurs: Startup Killer, the Cost of Customer Acquisition (no date on the page)forentrepreneurs.com · checked 03/11/2026
- David Skok, forEntrepreneurs: SaaS Metrics 2.0, LTV to CAC guidelines (subscription companies)forentrepreneurs.com · checked 03/11/2026
- David Skok, forEntrepreneurs: SaaS Metrics 2.0, detailed definitionsforentrepreneurs.com · checked 03/11/2026
- David Skok, forEntrepreneurs: What's your TRUE customer lifetime value (LTV)? DCF provides the answerforentrepreneurs.com · checked 03/11/2026
- Fader, Hardie: What's Wrong With This CLV Formula? (note, December 2014)brucehardie.com · checked 03/11/2026
- Fader, Hardie: Average Lifetime is not 1/Churn Rate (note, July 2026)brucehardie.com · checked 03/11/2026
- Fader, Hardie: Customer-Base Valuation in a Contractual Setting: The Perils of Ignoring Heterogeneity, Marketing Science, 2010faculty.wharton.upenn.edu · checked 03/11/2026
- Shopify: Customer lifetime value, formulas and cohorts (vendor blog)shopify.com · checked 03/11/2026
- ChartMogul: LTV definition and limitations (vendor documentation)chartmogul.com · checked 03/11/2026
- OpenStax: Principles of Managerial Accounting, contribution marginopenstax.org · checked 03/11/2026
- Yandex Advertising: Cohort analysis, definitions of LTV, CAC and cohort (in Russian, updated 23 Sep 2024)b2b.yandex.ru · checked 03/11/2026
- Yandex Advertising: New Metrica retention reports (27 Aug 2026, in Russian)b2b.yandex.ru · checked 03/11/2026
- Yandex Metrica: Importing offline data (CRM orders with revenue and cost)yandex.com · checked 03/11/2026
- Yandex Direct: Revenue and conversion valueyandex.com · checked 03/11/2026
- Google Ads: About conversion valuessupport.google.com · checked 03/11/2026
- Google Ads: Customer lifecycle goalssupport.google.com · checked 03/11/2026
- Google Analytics 4: Cohort explorationsupport.google.com · checked 03/11/2026
- Google Analytics (former Universal Analytics): Lifetime Value report, deprecated 29 Jan 2024support.google.com · checked 03/11/2026