Unit economics: LTV, CAC, churn, and NRR
Does each customer make you money? Learn gross margin, CAC, LTV, payback, churn, and the retention numbers that investors read first.
In this lesson you will learn
- How to calculate gross margin, CAC, LTV, and CAC payback
- Why LTV must come from gross profit, not from revenue
- The difference between logo churn and revenue churn, and how monthly churn compounds
- How to calculate GRR and NRR, and what an NRR of more than 100% means
A startup can grow fast and lose money on each customer. Then growth makes the losses bigger, not smaller. Unit economics answers two questions about one customer. How much does the customer give you over its lifetime? How much does it cost to get that customer?
The answers tell you if more sales make the company stronger or weaker. Investors read these numbers early, because a model that loses money on each customer cannot correct the loss with more customers.
Predict first
Each customer pays $500 a month. The gross margin is 80%, and 2% of customers leave each month. How much gross profit does one customer give you over its lifetime, on average?
Type a guess before you continue. A guess first, even a wrong one, makes the answer easier to remember.
One customer gives $500 × 80% = $400 of gross profit a month. If 2% of customers leave each month, the average customer stays 1 ÷ 0.02 = 50 months. Thus the lifetime gross profit is $400 × 50 = $20,000. That number is the LTV.
Gross margin
Gross margin is the part of each revenue dollar that stays after you pay the direct costs to deliver the product. The direct costs are the cost of revenue: hosting, AI inference (the cost to run the model), payment fees, and customer support.
Gross margin
= (Revenue − Cost of revenue)
÷ Revenue
| Software company | AI app | |
|---|---|---|
| Revenue | $1,000,000 | $1,000,000 |
| Cost of revenue | $200,000 | $550,000 (inference $500,000, hosting and support $50,000) |
| Gross profit | $800,000 | $450,000 |
| Gross margin | 80% | 45% |
a16z wrote in 2015 that software companies “should have very high gross margins, in the 80%-90% range.” AI products pay for compute each time a customer uses them, so their margins are lower. ICONIQ reported an average gross margin of 45% for AI products in 2025. The lesson What investors expect shows how investors look at that margin.
CAC: the cost to get one customer
CAC (customer acquisition cost) is the cost to get one new customer. The sales and marketing costs include the salaries of the sales and marketing team:
CAC = Sales and marketing costs
÷ New customers
= $60,000 ÷ 10
= $6,000
Two versions exist. Blended CAC divides by all new customers, which includes customers who came free through recommendations. Paid CAC counts only the customers from paid channels. a16z writes that CAC “should be the full cost of acquiring users,” and that investors give paid CAC more weight.
Churn
Churn is the part of your customers or revenue that you lose in a period. Two kinds of churn can tell different stories:
- Logo churn counts customers. Each customer counts the same, small or large.
- Revenue churn counts recurring revenue that leaves, from cancellations and downgrades.
At the start of a year, 100 customers pay $100K of MRR in total. During the year, 10 small customers leave, and they paid $5K MRR in total. Logo churn is 10%, but revenue churn is only 5%.
Monthly churn compounds, like growth. Each month, the rate applies to the customers that are still with you:
Customers left after a year
= (1 − Monthly churn)^12
| Monthly churn | Customers left after one year |
|---|---|
| 1% | 88.6% |
| 2% | 78.5% |
| 3% | 69.4% |
| 5% | 54.0% |
| 7% | 41.9% |
A 5% monthly churn sounds small. In one year, the company loses approximately 46% of its customers.
LTV: what one customer is worth
LTV (lifetime value) is the gross profit that one customer gives you during all of the time that it stays. The usual simple formula:
LTV = Monthly revenue per customer
× Gross margin ÷ Monthly churn
= $500 × 80% ÷ 2%
= $20,000
Why divide by churn? If 2% of customers leave each month, the average customer stays 1 ÷ 0.02 = 50 months. Thus LTV = $400 of monthly gross profit × 50 months.
a16z is more strict than this formula. It uses contribution margin, which also subtracts other variable costs. The two sources, a16z and David Skok, calculate LTV from a margin, not from revenue.
LTV:CAC and CAC payback
The LTV:CAC ratio shows how many dollars of lifetime gross profit you get for each dollar that you spend to get a customer. In 2010, David Skok gave this rule: “LTV must be at least 3x greater than CAC.”
CAC payback is the number of months of gross profit that it takes to earn back the CAC. Skok’s first rule was less than 12 months. He later wrote that “many of the best SaaS businesses are able to recover their CAC in 5-7 months.”
Worked example · step 1 of 5
The unit economics of one customer
- 1
Gross profit each month
The customer pays $500 a month. The company keeps 80% after direct costs.
Gross profit = $500 × 80% = $400 a month
Use the calculator to see which input moves LTV the most. Start with the numbers from the worked example. Double the churn to 4%. Then set the churn back to 2%. Cut the margin to 40%. Last, select “Revenue (a mistake)”. Compare the LTV.
Calculator
Unit economics: what one customer is worth
Set the price, the gross margin, the churn, and the cost to get one customer.
- Gross profit
- $400
- Each month
- Average lifetime
- 50 months
- 1 ÷ 2%
- LTV
- $20K
- LTV:CAC
- 3.33x
- Rule of thumb: 3x or more
- CAC payback
- 15 months
- Rule of thumb: 12 months or less
- After one year
- 78%
- Of customers stay
Gross profit from one customer, total so far
- A customer who stays
- The average customer
- CAC
Each customer brings $400 of gross profit a month and stays 50 months on average, so LTV is $20K. That is 3.33x the $6K CAC, which meets the 3x rule of thumb.
The company earns back the CAC in 15 months, slower than the 12-month rule of thumb. The average customer pays back later, because some customers leave first.
The chart has three lines. The green line is a customer who never leaves: it crosses the CAC at the payback month. The blue line is the average customer. Some customers leave, so the blue line bends, and over many years it gets close to the LTV. If the LTV is less than the CAC, the blue line never crosses the CAC line, and the average customer never pays back.
The table changes one input at a time. The base case is $500 a month, an 80% gross margin, 2% monthly churn, and a $6,000 CAC.
| Change | LTV | LTV:CAC | Payback |
|---|---|---|---|
| None (base case) | $20,000 | 3.33 | 15 months |
| Churn of 4% | $10,000 | 1.67 | 15 months |
| Margin of 40% | $10,000 | 1.67 | 30 months |
| LTV from revenue (a mistake) | $25,000 | 4.17 | 12 months |
Work it out
The price is $400 a month, the gross margin is 50%, monthly churn is 2.5%, and CAC is $4,000. What is the LTV:CAC ratio?
LTV = $400 × 50% ÷ 2.5% = $200 ÷ 0.025 = $8,000. LTV:CAC = $8,000 ÷ $4,000 = 2.0. That is less than the 3x rule. If you use revenue, you get $16,000 and 4.0, which hides the problem.
Can I trust an LTV from six months of data?
Be careful. The formula 1 ÷ churn uses one churn rate for all of the years in the future. A short history cannot show that churn stays at that rate. When the history is short, a16z recommends an LTV for 12 months and for 24 months. That is the gross profit that a customer gives in its first year or its first two years.
GRR and NRR: what last year’s customers pay now
The retention numbers compare what last year’s customers pay now with what the same customers paid one year ago. New customers do not count. The two numbers count different changes:
- GRR (gross revenue retention) counts only decreases: cancellations and downgrades. Thus GRR cannot be more than 100%.
- NRR (net revenue retention) also counts upgrades and price increases, the expansion revenue. Thus NRR can be more than 100%.
SaaS Capital uses these formulas:
GRR = (Start − Cancellations
− Downgrades) ÷ Start
= ($100K − $5K − $3K) ÷ $100K
= 92%
NRR = (Start − Cancellations
− Downgrades + Upgrades)
÷ Start
= ($100K − $5K − $3K + $20K)
÷ $100K
= 112%
An NRR of more than 100% shows that last year’s customers alone make the business bigger. Some investors call this “negative churn”. At 112% NRR, a company with $10M of ARR starts the next year at $11.2M before new sales. At 90% NRR, the same company must sell $1M of new ARR only to stay flat.
Calculator
GRR and NRR: the leaky bucket
Take the customers you had one year ago. Set what they cancelled, what they downgraded, and what they added.
- GRR
- 92%
- Losses only
- NRR
- 112%
- Losses and upgrades
- ARR today
- $1.12M
- From these customers
- To stay flat
- $0
- New ARR to sell
From last year's ARR to this year's
- A year ago
- $1M
- Cancelled
- $50K
- Downgraded
- $30K
- Upgraded
- $200K
- Today
- $1.12M
These customers kept 92% of their revenue before upgrades: that is the GRR. With upgrades, they pay 112% of what they paid a year ago: that is the NRR.
Last year's customers alone grow ARR by 12%, with no new sales. Each new customer adds more growth.
Work it out
A year ago, your customers paid $500K of ARR. Since then, they cancelled $40K, downgraded $10K, and upgraded $75K. What is the NRR?
NRR = ($500K − $40K − $10K + $75K) ÷ $500K = $525K ÷ $500K = 105%. GRR = ($500K − $40K − $10K) ÷ $500K = 90%.
Quiz · question 1 of 5
Test what you learned
A SaaS company reports 125% NRR. Which statement must be correct?
Key takeaways
- 1.Gross margin = (revenue − cost of revenue) ÷ revenue. Traditional software keeps 80% to 90%, and AI products keep less.
- 2.LTV = monthly revenue per customer × gross margin ÷ monthly churn. A common rule of thumb asks for LTV of at least 3 times CAC.
- 3.CAC payback = CAC ÷ monthly gross profit per customer. A common rule of thumb asks for 12 months or less.
- 4.Monthly churn compounds. At 5% a month, approximately 46% of the customers leave in one year.
- 5.GRR counts only losses and cannot be more than 100%. NRR adds upgrades. An NRR of more than 100% means that last year’s customers alone grow the company.
Sources
- 1.16 Startup Metrics. Andreessen Horowitz, August 21, 2015.
- 2.16 More Startup Metrics. Andreessen Horowitz, September 23, 2015.
- 3.SaaS Metrics – A Guide to Measuring and Improving What Matters. David Skok, forEntrepreneurs, February 17, 2010.
- 4.SaaS Metrics 2.0. David Skok, forEntrepreneurs, Approximately 2013–2014.
- 5.Research Brief 32: 2025 B2B SaaS Retention Benchmarks. SaaS Capital, September 2025.
- 6.The SaaS Retention Report: The AI churn wave. ChartMogul and Growth Unhinged, December 10, 2025.
- 7.2026 State of AI Report: The Builder's Economy. ICONIQ Growth, July 2026.