Not legal advice: The worked example below uses stated, hypothetical inputs (price, margin, churn, acquisition cost), not your data. Replace them with your own before you rely on any ratio.
Open any article on customer lifetime value and you will meet the same advice: your lifetime value (LTV) should be at least three times your customer acquisition cost (CAC). It is repeated as though it were arithmetic. It is not. It is a rule of thumb that came out of subscription-software investing, and it travels badly once you leave that world.
We went back to the source and to the academic work on how customers actually stay and leave. The short version: the 3:1 figure is best known from David Skok's essay on SaaS metrics, where it is explicitly a guideline. The one-line formula usually used to compute LTV assumes every customer churns at the same rate, and in the one real dataset we could open, that assumption was wrong by a factor of more than forty at year twelve. And long-standing customers were not reliably the profitable ones. For a small business the sturdier question is simpler: how many months until a customer has paid back what it cost to win them?
Key takeaways
- The "3:1" benchmark is best known from David Skok's "SaaS Metrics 2.0" essay, which says the best SaaS businesses are "higher than 3, sometimes as high as 7 or 8" and stresses "these are only guidelines." It is a software-subscription benchmark, not a universal law.
- The same essay puts the best SaaS businesses at recovering CAC in 5 to 7 months and calls profitability "anemic" beyond 12. Payback time is the second half of that guidance and, in our view, the half that travels best outside SaaS.
- The common LTV shortcut (monthly gross profit divided by churn) assumes one constant churn rate. In Fader and Hardie's 2007 data, a customer segment kept 63.1% of its customers after year one but 94.5% year over year by year twelve, because the leavers go first.
- If you project that first-year rate forward, you predict about 0.4% of the group still active at year twelve. The actual figure was 17.3%.
- Reinartz and Kumar (HBR, 2002) tracked more than 16,000 customers at four companies and found the link between how long customers stay and how much profit they bring was weak to moderate (correlations of 0.20 to 0.45).
- Build LTV from gross profit, not revenue, and take Israeli VAT (18% since 1 January 2025) out of revenue before you do. Then compare cohorts at 3, 6 and 12 months against CAC by channel, and treat any "lifetime" number as a projection to be tested.
Where does the 3:1 LTV:CAC rule come from?
From one essay, and it says less than the people quoting it. David Skok's "SaaS Metrics 2.0" on his For Entrepreneurs site lays out two unit-economics guidelines for subscription software. The first is the ratio: "The best SaaS businesses have a LTV to CAC ratio that is higher than 3, sometimes as high as 7 or 8." The second is time: "many of the best SaaS businesses are able to recover their CAC in 5-7 months," and profitability is "anemic if the time to recover CAC extends beyond 12 months."
Three caveats sit in the same essay and rarely travel with the number. Skok writes that "these are only guidelines, there are always situations where it makes sense to break them"; that "many healthy SaaS businesses don't meet the guidelines in the early days"; and that bigger companies can accept longer payback because they have cheap capital, while startups usually find capital expensive. The essay also does not spell out the formulas: it defines LTV and CAC in a sentence and sends the reader to a separate definitions page. The essay is undated, but it carries a May 2016 update note and describes itself as a rewrite of an older post.
| Widely repeated version | What the source says | Status |
|---|---|---|
| "LTV must be 3x CAC" | "higher than 3, sometimes as high as 7 or 8" for the best SaaS businesses, offered as a guideline | Real, but a SaaS guideline |
| "Payback must be under 12 months" | Best SaaS recover CAC in 5 to 7 months; beyond 12 months profitability is "anemic" | Real, SaaS |
| "It applies to every business" | Not claimed; the essay is written for SaaS and says to break the guidelines where it makes sense | Not in the source |
We read Skok's essay directly. Where we paraphrase, it is marked; the quotations are word for word. We cannot see how many other businesses (clinics, studios, shops) meet these thresholds, because the essay does not say.
What goes wrong with the one-line LTV formula?
The shortcut most guides teach is: average monthly gross profit per customer, divided by the monthly churn rate. A customer who yields ₪150 a month with 5% monthly churn is "worth" ₪3,000. The formula quietly assumes every customer has the same, unchanging chance of leaving each month. Real customer bases do not behave that way.
Peter Fader and Bruce Hardie, two marketing-analytics professors, showed this with real subscription data in "How to Project Customer Retention" (Journal of Interactive Marketing, 2007). In the "Regular" segment of an unnamed subscription business, 63.1% of customers were still there after one year, 46.8% after two, and 17.3% after twelve. Look at the year-over-year retention hiding in those numbers: 63% in year one, then 74%, 82%, 85%, 89%, 91% and 92%. By year twelve it was 94.5%.
Nobody got more loyal. The authors' explanation is that each customer keeps a constant, unknown chance of leaving, and chances differ across customers, so "the high-churn customers drop out early in the observation period, with the remaining customers having lower churn probabilities." They call it a sorting effect. The consequence for the shortcut is large: an analyst who takes year one's 63.1% and projects it forward predicts 0.4% of the group still active at year twelve (0.631 to the twelfth power, our arithmetic). The real number was 17.3%.
17.3% vs 0.4%
of the group still active at year 12: what actually happened versus what projecting year one's retention rate forward would predict.
Source: Fader & Hardie (2007), Table 1; the 0.4% is our arithmetic (0.631^12)
The data come from a contractual, subscription-type business where cancellations are visible, and the authors do not name the company. Treat the shape (retention rising as the base sorts itself) as the lesson, not the exact percentages as a benchmark for your business.
Do long-standing customers always make more money?
The loyalty story behind lifetime value says yes: the longer someone stays, the more they are worth. Werner Reinartz and V. Kumar tested it in "The Mismanagement of Customer Loyalty" (Harvard Business Review, July 2002) on customer databases from four companies: a US high-tech corporate service provider, a US mail-order company, a French retail food business and a German direct brokerage. The data covered more than 16,000 individual and corporate customers over four years.
They found the relationship much weaker and subtler than loyalty advocates claim. The correlation between how long a customer had stayed and the profit they brought was weak to moderate: 0.45 at the grocery retailer, 0.30 at the corporate service provider, 0.29 at the brokerage and 0.20 at the mail-order company. They found "little or no evidence" that steady customers were cheaper to serve, less price sensitive, or especially good at bringing in new business. In the service provider's own data, about half of the customers who had bought regularly for at least two years barely generated a profit, and about half of the most profitable customers were "blow-ins" who bought a lot of high-margin products in a short time and vanished.
In their four-way split, between 15% and 20% of customers at each company were short-term but highly profitable (the authors' "butterflies"), and between 15% and 21% were long-term but only marginally profitable (the "barnacles"). A single average lifetime blurs both groups into one figure.
This is a 2002 study of four large companies, not small businesses. It shows that tenure and profit can come apart, not how large the gap is for a clinic or a studio.
What changes when your customers never "cancel"?
Most small businesses, including shops, clinics, studios and course sellers, do not sell contracts with a visible end date. A customer who last bought in March might be gone for good or might be back in November, and you cannot see which. Fader and Hardie draw the line explicitly: in a non-contractual business, "the time at which a customer becomes inactive, and the likelihood that it has occurred at all, must be inferred from the transaction history." That is why a monthly churn rate is not simply there to be read off a dashboard.
Academic models exist for this case. The Pareto/NBD model of Schmittlein, Morrison and Colombo (Management Science, 1987) estimates each customer's likelihood of still being "alive" from purchase timing, and Fader, Hardie and Lee's 2005 BG/NBD model is a much easier-to-implement alternative that gives very similar results. For a small business without a data scientist, the practical takeaway is more modest: stop pretending a fixed lifetime exists, and measure what each group of customers has actually paid so far.
How far apart can one business's ratio be?
Very far. To show it, we built a deliberately simple hypothetical: a studio that sells a ₪250 monthly membership (VAT included). Israeli VAT has been 18% since 1 January 2025, so revenue is ₪250 / 1.18 = ₪211.86 a month. At a 70% gross margin (our assumption) that is ₪148.31 of gross profit a month. Winning each member costs ₪600.
Now give the customers two hidden types that you cannot tell apart at signup: half leave at 20% a month, half at 2% a month. The blended base loses 11% in the first month and then retains better and better, exactly the pattern in the Fader and Hardie data. Here is the same studio, read four ways.
| How you read it | Paid months per customer | LTV (gross profit) | LTV:CAC |
|---|---|---|---|
| One-line formula on first-month churn (11%) | 9.1 | ₪1,348 | 2.2 : 1 |
| Actually collected in the first 12 months | 7.7 | ₪1,143 | 1.9 : 1 |
| Collected over the first 24 months | 12.1 | ₪1,794 | 3.0 : 1 |
| Full lifetime, with both hidden types | 27.5 | ₪4,078 | 6.8 : 1 |
The same studio recovers its ₪600 by the fifth monthly payment (₪605 collected), at the fast end of Skok's 5-to-7-month guideline, while a quick reading of the one-line formula says it fails 3:1. And 25 of the 27.5 expected lifetime months, 91%, come from the half of customers who rarely leave, a group you can only confirm by waiting.
Which inputs do people forget?
- VAT. If you quote prices including VAT, take it out. Israeli VAT is 18% since 1 January 2025 (it was 17% before), so ₪250 of customer money is ₪211.86 of your revenue.
- Gross margin, not revenue. LTV that counts revenue overstates what a customer is worth by the cost of delivering to them. The standard textbook CLV expression, as Fader and Hardie write it, multiplies a per-period margin by the survival curve and discounts later periods.
- Discounting. A shekel next year is worth less than one today, especially if you borrow to fund growth. The same textbook expression divides each period by (1 + discount rate) to the power of the period; Skok notes that startups typically find capital expensive.
- Refunds, chargebacks and failed payments. They come out of what you collected.
- Segments and channels. Skok's essay recommends computing CAC, LTV:CAC and payback for each segment separately, and notes that different lead sources carry different costs. A blended average can hide a channel that loses money.
- Time. A cohort acquired six months ago has six months of history. Calling its projection "LTV" invites the mistake in the studio table above.
What should a small business track instead?
Our suggestion, not a study result: use the 3:1 ratio as a sanity check at most, and run the business on payback and cohorts, which you can observe rather than assume. Skok's second guideline, time to recover CAC, fits that approach.
- 1Group customers into cohorts by the month they first paid and, if you can, by the channel that brought them.
- 2For each cohort, add up gross profit actually collected per customer acquired after 3, 6 and 12 months. These are facts, not projections.
- 3Work out CAC per channel from everything you spent winning those customers in the same period: ad spend plus the paid time and tools behind it.
- 4Mark the payback month, the first month the cohort's cumulative gross profit per customer passes its CAC. For software, Skok's guideline is 5 to 7 months at best and trouble beyond 12; the right line for you depends on your margin and how much cash you can lock up.
- 5Only extend to a lifetime figure when a cohort has at least a year of history, use a method that allows retention to rise over time, and discount it.
- 6Split by customer type. Tenure alone told Reinartz and Kumar little about profit, so keep an eye on the short-term big spenders and the long-term low spenders separately.
| Metric | What it tells you | Can you observe it today? |
|---|---|---|
| CAC payback month (by channel) | How long cash is tied up winning a customer | Yes, per cohort |
| Gross profit per customer at 12 months | What a cohort has actually paid back | Yes, once 12 months have passed |
| LTV:CAC from a lifetime projection | A forecast that rests on assumed retention | No, it is an estimate |
A longer horizon is not automatically better. If your market changes fast, a projection built on old cohorts can mislead you more than a short, observed window.
How do you run this in oToK?
oToK is built around keeping every payment and sale on the customer's record, which is what cohort analysis needs. Payments are recorded on each contact (one-time, recurring and installments), with revenue analytics on the contact showing LTV, MRR/ARR and trends over time, and sales you log (multi-item, with the buyer and recipient kept separate) roll up to the contact as well. Because contacts, deals and pipelines are unlimited on every plan, the customer history you build is not capped by your list size.
To get CAC by channel, attribution (available on Growth and up) joins ad spend from your connected Meta Ads and Google Ads accounts to the revenue each channel produced, with ROI and ROAS by channel, campaign and ad, including Click-to-WhatsApp ads. The report builder (saved reports are available on Growth and up) lets you group contacts by when they first paid and compare how much each group has paid since, and customer segments such as recent high spenders can be built from the same records. Segmenting by recency, frequency and value is covered in our RFM guide linked below.
- Record every payment and sale on the contact so history accumulates by itself
- Read LTV and MRR/ARR on the contact, not in a spreadsheet
- Compare channels on revenue, not just clicks, with attribution on Growth and up
- Cut a segment of recent high spenders for a follow-up, and a separate one for long-time low spenders
Frequently asked questions
- What is a good LTV:CAC ratio?
- The figure most people quote is 3:1 or higher, which comes from David Skok's SaaS Metrics 2.0 essay: the best SaaS businesses are "higher than 3, sometimes as high as 7 or 8," offered as a guideline that can be broken. It describes subscription software, so for another kind of business use it only as a rough check and rely on payback time and observed cohorts.
- Where does the 3:1 rule come from?
- It is best known from David Skok's essay on SaaS metrics. The essay states the guideline for subscription software businesses and adds that it is "only guidelines" with situations where it makes sense to break them. It does not present 3:1 as a law for all businesses.
- What is CAC payback period and what is a good one?
- CAC payback is how long it takes a customer's gross profit to repay the cost of acquiring them. Skok's guideline for SaaS is that many of the best recover CAC in 5 to 7 months, and profitability gets "anemic" beyond 12 months. For other businesses, the right threshold depends on your margin and how much cash you can tie up.
- How do I calculate customer lifetime value without subscriptions?
- Do not assume a lifetime. Take customers who first paid in the same month and add up the gross profit they have actually generated after 3, 6 and 12 months. In a business with no visible cancellation, whether a quiet customer is gone must be inferred from purchase history, so observed cohorts are safer than a formula.
- Should I use revenue or gross profit in LTV?
- Gross profit. Revenue ignores what it costs to deliver the product or service, so it overstates what a customer is worth. Also remove VAT before you start; Israeli VAT has been 18% since 1 January 2025.
- Is the formula LTV = monthly profit / churn reliable?
- Only when every customer has the same, unchanging chance of leaving. In Fader and Hardie's 2007 data, retention rose from 63.1% in year one to 94.5% in year twelve, and extending the first-year rate would have predicted 0.4% of the group still active at year twelve against an actual 17.3%. Use it as a rough floor, not a forecast.
- Are loyal customers always more profitable?
- No. Reinartz and Kumar's 2002 study of four companies and more than 16,000 customers found only weak-to-moderate correlations between tenure and profit (0.20 to 0.45), and found little or no evidence that long-standing customers were cheaper to serve or less price sensitive. It is a study of large companies, but it is a good reason to measure profit by customer, not just tenure.



