Customer Lifetime Value: The Cohort Math Most Owners Get Wrong

The formula every business owner learns for customer lifetime value is simple. Average revenue per customer times gross margin divided by churn rate. Plug in three numbers, get LTV. Easy.

That formula is also wrong most of the time, because the inputs are averages and customer behavior is not average. Real LTV lives at the cohort level, and the difference between the simple number and the cohort number is usually 2x or more.

The Problem With Average LTV

Take a business with average customer revenue of $200 a month, 65 percent gross margin, and 5 percent monthly churn. The textbook LTV is $200 times 0.65 divided by 0.05, which equals $2,600. That number gets stamped on every marketing budget meeting for the next year.

Now run the actual math. Pull a cohort of 100 customers who joined in January 2024. Track them month by month. By month 6, 38 had churned, not the 30 the average predicted. By month 12, 51 had churned. By month 24, only 27 were still around. Real LTV for that cohort came out to $1,840, not $2,600. Acquisition decisions made on the wrong number led to overspending on marketing for 18 months.

According to research from the Stanford Graduate School of Business, businesses that segment LTV by cohort and channel make customer acquisition decisions that produce 2 to 4 times higher returns on marketing spend than those using a single average LTV. The difference is real money. The simple formula does not just round wrong, it systematically biases toward overspending on customer acquisition.

Why Average Churn Lies

The reason cohort math diverges from average math is survivorship bias. Average churn rate measures every customer in the business at one point in time. That sample is dominated by long-tenured customers, because the bad-fit customers churned years ago and dropped out of the average.

The customers you are acquiring this month are not the long-tenured cohort. They are a fresh batch with their own churn curve. The first 6 months of any cohort have churn rates 2 to 3 times higher than the long-run average, because that is when bad-fit customers self-select out. Using average churn to forecast a new cohort underestimates early churn and overestimates LTV.

The fix is to model LTV with a survival curve, not a single churn rate. Different cohorts have different curves. Customers acquired through paid ads usually churn faster than referrals. Customers on a discount usually churn faster than full-price. Seasonal cohorts behave differently from steady-state cohorts. Average everything together and you get a number that is technically correct and operationally useless.

The CFO Perspective

"The simple LTV formula tells you what an average customer is worth. The cohort model tells you what the next customer you acquire is going to be worth. Those are two different questions, and only the second one matters for budgeting." Peter Xia, CPA

I worked with a $2.8M revenue subscription business last year that had been running on a $4,200 average LTV number for 3 years. We rebuilt the model with cohort math segmented by acquisition channel. Referral cohorts had LTV of $5,800. Paid social cohorts had LTV of $2,100. Webinar cohorts had LTV of $4,900. The average was $4,200 because referrals and webinars dragged paid social up. She had been pouring 55 percent of marketing spend into the worst-performing channel because the average looked fine. We cut paid social spend by 60 percent, redirected to webinar production and referral incentives, and net new revenue grew 22 percent over the next 6 months on lower marketing budget.

How to Build a Real Cohort LTV Model

1. Pull every customer from your billing system with their start date, monthly revenue, and current status. Group them by acquisition month. Each month is a cohort.
2. For each cohort, calculate cumulative gross margin per starting customer at month 1, 3, 6, 12, 18, and 24. This is the cohort's LTV curve at each milestone.
3. Segment cohorts by acquisition channel. Referral, paid search, paid social, content, partnerships. Each channel produces customers with different churn behavior. Calculate the LTV curve for each channel separately.
4. Compare the LTV curve to the channel's customer acquisition cost. The ratio of 12-month LTV to CAC is the operating number. The 24-month ratio is the long-term number. Both should be above 3 to 1 for a healthy channel.
5. Identify the worst channel by 12-month LTV to CAC. Cut spend on that channel by 50 to 80 percent and reallocate to the channel with the best ratio. Test for 90 days, measure, repeat.
6. Segment cohorts further by customer profile. Industry, business size, plan tier, geography. The customer-level segmentation usually surfaces a profitable niche hidden inside an average channel and an unprofitable niche hidden inside a profitable one.
7. Update the cohort model quarterly. Customer behavior shifts when pricing changes, when product changes, when the economy changes. The model is not a one-time build, it is a rolling read on the business.

The Bottom Line

Average LTV is a comfortable number that hides expensive mistakes. Cohort LTV by channel is the real picture. The difference between the two is often the difference between marketing spend that compounds and marketing spend that drains the line of credit. If you want the cohort LTV spreadsheet I use with my CFO clients, book a free call at peterxiacpa.com/book.

Next step: run your numbers through the free CFO scorecard.