Revenue is up 20% year over year. Every month beats the one before it. The chart in your investor update points up and to the right.
Now answer a different question. Of the customers who bought from you 6 months ago, how many bought again?
Most consumer brands can’t answer that because customer retention isn’t anywhere in the financial reporting they get every month.
What Is Cohort Analysis? What the P&L Doesn’t Tell You
That’s the question cohort analysis answers. A monthly revenue line on the P&L adds new customers and returning customers together and reports one number. Two businesses can produce identical revenue. One keeps its customers. The other replaces them every month and pays to gain those new customers.
Let’s say one company acquires 4,000 customers a month and keeps a meaningful share of them, so each month builds on a base that stays. The other has to acquire 6,000 customers a month to keep up, watches nearly all of them leave, and buys 6,000 more. Same revenue. Same growth rate. But at a much steeper cost.
When acquisition costs rise, or a channel gets more expensive, maintaining the same revenue growth requires higher spend.
Cohort analysis is the reporting that separates those two companies. It groups customers by the month they first bought and tracks each group forward as its own unit, since a customer you acquired 2 years ago who still buys is very different from the one you acquired last week.
Cohort Analysis Example: What a Cohort Table Looks Like
Take this illustrative cohort table for a CPG brand:
| Cohort | Customers | Month of First Purchase | Month 1 | Month 2 | Month 3 | Month 4 | Month 5 | Month 6 |
|---|---|---|---|---|---|---|---|---|
| January | 4,000 | 100% | 17% | 13% | 11% | 9% | 8% | 7% |
| February | 3,500 | 100% | 16% | 13% | 11% | 8% | 7% | |
| March | 6,500 | 100% | 9% | 6% | 5% | 5% | ||
| April | 4,000 | 100% | 16% | 12% | 11% | |||
| May | 4,500 | 100% | 15% | 12% |
January’s customers are down to 7% repeating monthly by month 6. If that holds for the next 5 months, that adds up to one more order for each customer, so the average January customer places two orders over about a year and generates about $220 at a $110 order value. That $220 is where the math starts.
Then there’s March. It brought in 6,500 customers, the best acquisition month on the table, and those customers repeat at roughly half the rate of the months on either side of it. March revenue was the standout, but retention was the lowest.
How Much Can You Spend to Acquire a Customer?
$220 is what a January customer spends. It isn’t what the business keeps. Product, freight, payment fees and returns come out first, and depending on your margins you might be left with something closer to $135.
That $135 is what the business actually has to work with. It has to cover acquiring the customer, your overhead, and whatever profit you’re targeting, so acquisition only ever gets a share of it. Deciding how big that share can be is the whole question.
Put that next to what you paid and January reads like this. You spent $85 to acquire a customer who returns about $135, which leaves $50 toward overhead and profit. Across 4,000 customers, that’s $200,000 from one month of acquisition. That’s about 1.6x the $85 over the first year. Continuing to focus on all levers to either decrease CAC or increase retention over time will result in considerably higher marketing efficiency, typically defined as an LTV:CAC ratio.
March is where it turns. Those customers settle near 5% instead of 7%, so they place about 1.6 orders each instead of two, and the discounting that brought them in pulls the margin closer to 50%. They also cost more to get, around $100, because spend scaled hard to hit that volume. Run the same math and roughly $88 comes back against $100 spent. The average customer in that cohort lost $12, about $78,000 across the cohort, while the revenue chart showed the best month of the year.
One more thing the table hides. You pay the $85 up front and only about half the $135 comes back with that first order. The rest arrives in pieces across the year, and you bought their inventory before any of them showed up.
What Skipping Cohort Analysis Costs
- You can’t price acquisition. Without a number like the $135 above, every decision about what to pay for a customer is a guess. Brands that guess low leave growth on the table. Brands that guess high scale a loss.
- You buy inventory for customers who aren’t coming back. This is the expensive one. A company that misjudges retention places a purchase order against demand that doesn’t exist. The cash sits in product in a warehouse instead of funding growth.
- You double down blind. March brought in 6,500 customers against a normal 4,000, so it reads as the best month of the year and the budget follows it. Those customers repeat at half the rate of the months around it. Without the cohort view, you keep scaling it without seeing that it loses money.
- Investors will ask for it. Cohorts come up early in diligence, and a blended lifetime value with no cohort table behind it invites a longer conversation. Brands that have the view ready spend that time talking about the business instead of assembling the analysis.
The Decisions Cohort Analysis Changes
What you can afford to pay for a customer. How often they come back, times what they spend, less product, freight, payment fees and returns, gives you what the business has to work with. Acquisition gets a share of it, and every acquisition decision sits inside that share.
Whether a discount was worth running. A promotion that drives a spike in first orders and a collapse in repeat behavior can cost more than it earns. You only see it by tracking how that month’s cohort behaves.
Which channel actually works. Two channels can deliver customers at the same cost and produce completely different lifetime value (LTV). Cohorts split by channel help you understand where to invest.
The Question Underneath All of This
Every month you spend money to acquire new customers. Whether that's a good trade depends on what those customers do over the next year, and the only way to know this is from cohort analysis.
What a customer is worth over time determines your acquisition budget, your inventory buys, your forecast, and what a buyer or an investor thinks the business is worth. It’s a knowable number, and the brands that measure it make better decisions than the ones guessing.
Crew’s FP&A team builds cohort analysis and KPI dashboards for startup consumer brands.