The LTV to CAC ratio that every deck quotes at three to one traces back to a single blog post. David Skok published SaaS Metrics 2.0 on 16 January 2013, describing what he had seen in the strongest subscription software companies. No dataset came with it. No sample size, no cited study, and no ecommerce data at all. Skok called his own figures “only guidelines” and named situations where breaking them made sense. Thirteen years later the number gets quoted as a law, usually by somebody about to cut a budget.
This is for ecommerce operators being held to a benchmark whose source nobody in the room has read.
Where the LTV to CAC ratio rule came from
One post, one author, one industry.
The source is SaaS Metrics 2.0 by David Skok, published on 16 January 2013. It describes the strongest SaaS businesses as running above three, and sometimes as high as seven or eight. That is the entire provenance.
Read what is absent. No dataset sits behind the figure, no sample of companies is stated, and no study is cited. One experienced investor described a pattern he had seen.
Publishing that is perfectly reasonable. Presenting it as a benchmark is not, and the author never did.
Thirteen years of citation did the rest. Each repetition dropped a little more context, until the figure reached ecommerce decks with no author, no date and no industry attached to it.
What the 2013 post said, and what it did not
Skok described his figures as guidelines with exceptions, and wrote about situations where breaking them makes sense.
That qualification never survives the retelling. A guideline with stated exceptions turned into a pass mark, and a pass mark turned into a reason to cut spend on a channel that was working.
There is a second omission. The post says nothing about ecommerce, because it was not written about ecommerce.
Anybody applying it to a store selling physical goods has made a translation, and the translation is where the meaning goes. Two businesses can both report three to one and share almost nothing else.
Why a subscription rule travels badly to a store
Subscription revenue arrives monthly until somebody cancels, so churn decides everything.
Lifetime value in software is a projection off a churn rate, and that rate is measurable within a few months of launch. The model has a shape, and the shape is reliable enough to bid against.
Ecommerce has no churn rate. It has a repeat purchase rate, and a customer who has not ordered for nine months has not cancelled anything. They may simply not need another one yet.
That difference matters for the top half of the calculation. Software lifetime value is a forecast with a known decay curve. Ecommerce lifetime value is a forecast with no curve at all, built from whatever repeat behaviour the store can evidence.
Same arithmetic, completely different confidence in the result.
The LTV to CAC ratio has two shaky inputs
Both halves are estimates, and the errors run in the same direction.
The top half needs gross margin, which is the input clients withhold most often. Across my ecommerce clients margin runs between 20% and 50%, so the same revenue produces wildly different profit. It also needs returns deducted, and returns are handled by my clients’ in-house teams. They never reach my reporting and never flow back into the ad platform.
The bottom half needs customer acquisition cost, which sounds simple and is not.
So the top half is usually too high and the bottom half is usually too low. An LTV to CAC ratio built from two optimistic estimates flatters the business twice, in the same direction, at the same time.
Which acquisition cost goes underneath
Here is a real account, one month, four defensible answers.
On a US flag and banner ecommerce client I run Meta sales campaigns. Across all six sales ad sets the cost per purchase was $67.24, on 64 purchases. Excluding one failed test it was $53.72, on 61 purchases. The best ad set produced purchases at $29.37, and the worst active one at $93.96.
Every one of those figures is true. Each describes the same client in the same period.
Pick the account average and the ratio looks one way. Pick the best ad set and it looks like a different company. Nobody quoting a single number to a board tells them which of the four they chose.
What an average acquisition cost hides
The failed test in that account is worth describing, because it inverts what most people expect.
One ad set targeted a neighbouring state. It spent roughly $1,027, returned 0.25 ROAS, and produced three purchases at about $341 each. It also had the highest reach of any ad set in the account at 17,086, and the most link clicks at 542.
Every traffic metric said it was the winner. Revenue said it was the only loser.
Blend that ad set into the account and the acquisition cost rises by nearly $14 a purchase. Remove it and the comparison improves, while nothing about the customers changed. Averages are a choice about what to include, and the choice is rarely stated out loud. Your LTV to CAC ratio moves on that decision alone, before a single buyer behaves any differently.
What the LTV to CAC ratio ignores completely
Time, and therefore cash.
Three to one collected over twenty-four months and three to one collected inside thirty days are the same ratio and two entirely different businesses. The first needs funding for two years. The second funds itself.
It is also unitless, which hides scale. Five to one on $2,000 a month of spend is usually a sign that you are harvesting the cheapest demand available and stopping there.
That client of mine has never disclosed gross margin, so I have never been able to produce this comparison for them at all. I can give you their cost per purchase to the cent. The other half of the calculation was never shared with me, and inventing it would be worse than leaving it blank.
When the LTV to CAC ratio is still worth calculating
Once a quarter, as a direction check, with both inputs written down beside it.
It answers one question well: is the gap between what a customer is worth and what they cost getting wider or narrower. Trend beats level here, because your own errors are at least consistent from quarter to quarter.
What it cannot do is give you a target, because no ecommerce benchmark with a disclosed sample exists. Triple Whale reported a median ROAS of 2.04 across its platform in 2024, with Meta making up 68.3% of tracked spend. Even that figure arrives with no stated sample or date range on the page carrying it, which tells you how thin the published evidence is.
Your own margin is a better target than anybody else’s ratio, and it is the only one you can verify.
What to change this week
Three things.
Write down both inputs before you write the result. State the margin you used, the window you used, and whether returns were deducted. A comparison without those three notes cannot be checked by anybody, including you in six months.
Then calculate acquisition cost at two levels: the account average and your best consistent ad set or campaign. The distance between them is the honest measure of how much room you have.
Then stop quoting three to one as a pass mark. Replace it with a first-order profit floor from your own gross margin, which is a number your finance team can confirm and your media buyer can bid to.
Working out what a lead or an order is worth covers the valuation method underneath the top half. What a purchase actually costs at volume gives two of my own figures with their date ranges attached. The profit view of the lifetime number covers the window and margin decisions in detail. Raising order size before cutting acquisition cost is the cheapest way to improve unit economics. Organic demand changes the arithmetic too, which is what my SEO work is for. If your numbers do not survive scrutiny, book a teardown.