LTV:CAC Is the Most Abused Number in Your Deck
A founder told me his unit economics were fine.
“Our lifetime value to customer acquisition cost ratio looks really healthy. So why are we still bleeding money?”
I asked him two questions. What is your CAC payback period? And are you measuring blended CAC or paid CAC?
He paused. “Isn’t all CAC the same?”
No. And that gap is where his money was going.
A healthy LTV:CAC ratio is not a measurement. It is two forecasts divided by each other. Both forecasts lean the same way, which is optimistic. Then the result gets checked against a rule of thumb almost nobody has read the source of. Three to one. Green light. Next slide.
My argument is simple. The ratio is most dangerous when it looks healthy, because of what it conceals: cohorts that get worse every month, and paid spend your organic traffic is quietly subsidising.
1. The man who wrote the 3:1 rule told you when to stop using it
David Skok set the 3:1 benchmark. It became gospel. Then he published a correction, and hardly anyone read it.
He wrote that he made “a significant mistake in not telling my readers when it would make sense to compute LTV and CAC”. His condition is clear. The numbers only mean something once you have a repeatable, scalable growth process. Before that, you are dividing noise by noise.
He also gave the line I now repeat to founders. LTV:CAC ratios are to be used, not believed.
Most companies quoting 3:1 fail his test. The founder closed the biggest deals personally. The cohorts are six months old. The churn rate is a number someone typed into a cell.
If your growth motion is not repeatable yet, the ratio is decoration.
2. Your LTV is a forecast, and you rounded it up
Four moves inflate the numerator. Most decks make all four:
- Revenue instead of gross profit, which counts money you never keep
- One blended churn rate stretched across a full projected lifetime
- Cohorts too young for the retention curve to have flattened
- Survivorship, where value is measured on the customers who stayed
Every one of these errors runs in the same direction. Early churn is always worse than late churn. So a young cohort, judged by a blended rate, looks better than it is. Strip out the people who already left, and it looks better still.
You do not get a noisy estimate. You get a confident one that is wrong on the high side.
3. A healthy ratio hides cohort decay
This is the failure mode nobody catches in time, because the average holds while the business rots underneath it.
Blue Apron is the clearest public case. A detailed teardown by customer value researcher Daniel McCarthy found that every new cohort generated about $7 less revenue over its next six months than the cohort before it. Meanwhile, 72% of customers churned by month six. His estimate: 70% of recent customers would never break even against what it cost to acquire them.
None of that shows up in a blended ratio. The old cohorts prop up the average. The new ones, the ones your next dollar buys, are worse than anything in the number you are quoting.
Key Takeaway: The average customer does not exist. Only cohorts exist. A ratio that blends them tells you about a business you used to have.
4. Blended CAC lets your organic traffic pay for your ads
Now the denominator, which is softer than the numerator.
Blended CAC divides total spend by total new customers. It mixes people who cost nothing with people who cost a lot. That average is fine for the profit and loss statement. It is useless for deciding whether to spend more, because the next customer will not be average. The next customer is bought at the margin, and the margin is always the expensive end.
It gets worse. Scaling paid spend does not just cost more per customer. It buys worse customers. Social Capital’s diligence framework shows cohorts acquired during paid acquisition spikes retaining materially worse than the cohorts around them. In aggregate reporting, that signal vanishes.
Then there is the part most teams never test. Some of what you paid for, you already had. eBay ran the experiment properly and switched off paid search across 68 US markets. The impact on sales was, in the researchers’ words, “indistinguishable and not significantly different than zero”. Nearly all the traffic came back through free channels. The ads were harvesting demand, not creating it, and the platform booked the credit anyway.
So your CAC is understated twice. Once by blending, once by attribution.
5. Run the number that is harder to lie with
CAC payback period is the honest version. It has a fixed clock, and it forces you to use gross profit rather than revenue. You cannot fix it with a friendlier lifetime assumption.
It also has real spread. Across 342 software companies in the 2026 Aleph and Benchmarkit benchmarks, median CAC payback was 16 months. The top quartile recovered in six months or fewer. The bottom quartile took 24 months or more. Two companies can both report 3:1 and sit at opposite ends of that range. One funds its own growth. The other needs a round.
Three changes, in order:
- Measure CAC by channel and by cohort, never blended alone
- Judge payback on gross profit, on marginal spend, not the average
- Test incrementality on your largest line item before you scale it
For a mobile gaming and publishing company, I built LTV prediction and cohort models that reached 90% forecast accuracy. The accuracy mattered less than the resolution. We measured acquisition cost down to the channel and the cohort, so we could see which cohorts were decaying while the blended number stayed calm.
We ran one more test that almost nobody runs. We checked payback against the cash cycle. Ad platforms bill on their schedule. Store revenue arrives on a different one. If your payback lands outside that window, you are funding growth with working capital, even at a ratio your investors would call healthy.
Final Thoughts: A ratio is a governor, not a verdict
The objection I get is that LTV:CAC is still directionally useful. Direction is the first thing it loses. When the numerator decays and the denominator is blended, both errors point the same way, and the number drifts confidently upward while the business gets harder.
Use it as a ceiling check, not a verdict. Then run the business on cohort payback, and on what your next marginal dollar actually buys.
If your ratio looks healthy and your bank balance disagrees, the ratio is not the thing to trust. Book a call, and we will look at the cohorts underneath it, or connect with me on LinkedIn.
A note before you close this tab. The fact that you read this far tells me something. You already sense that the way you’ve been thinking about growth might be incomplete. That instinct is worth following.
Mervyn Chua is a growth-transformation consultant helping founders and CEOs build the strategic clarity and systems to grow in an AI-first world. If this raises questions worth exploring for your brand, let’s talk.
