Your monthly P&L averages cohorts that should never be averaged

What the monthly P&L can’t tell you

A monthly P&L pools two different things in one column. There's the revenue and contribution from customers acquired this month, and there's the repeat contribution from every customer acquired before. The acquisition cost on the same statement is only for this month's customers. So the report divides this month's investment against a blend of this month's return and last year's.

This creates a predictable distortion. Cut acquisition spend and the P&L improves immediately, because the cost disappears while the older cohorts keep paying. Scale spend and it deteriorates immediately, because the cost arrives before the return. Teams that read the monthly statement as a performance score tend to under-invest in exactly the months when the cohort economics are best.

A cohort, worked

Take the January cohort: 4,000 new customers at a $60 CAC, so $240,000 of acquisition spend. First-order AOV is $85 with contribution before marketing of $34 per order (after COGS, shipping, fees, and fulfillment). Repeat orders carry $38 of contribution, since they're not discounted to acquire. Cumulative repeat orders per customer build as 0.20 by month 3, 0.38 by month 6, 0.65 by month 12, and flatten to about 1.10 by month 24.

Cumulative contribution per customer is $34 plus $38 times repeat orders:

• Month 0: $34. Cohort: 4,000 × $34 = $136,000, minus $240,000 spend = −$104,000.
• Month 3: $41.60 per customer, cohort at −$73,600.
• Month 6: $48.44, cohort at −$46,200.
• Month 12: $58.70, cohort at −$5,200.
• Month 15: about 0.74 repeat orders, $62.12, cohort at +$8,500.
• Month 24: $75.80, cohort at +$63,200, or 1.26x the original spend.

The shape matters as much as the endpoint. The cohort is $104,000 under water on day one and recovers roughly $5,000 to $7,000 a month after that, which means the first year is a slow climb back to even, not a series of profitable months with a delayed bonus. Any plan that assumes otherwise will misstate cash.

Payback happens around month 14 or 15. The month-12 return on acquisition is 0.98x, meaning the cohort has returned 98 cents per dollar after a year. Whether that's a good business depends on how long you can fund the gap, which is a cash question as much as a margin one.

Why the monthly view flips the decision

Now run the same cohort through the monthly statement. Suppose existing customers contribute $150,000 of repeat margin this month. With $240,000 of spend and $136,000 of first-order contribution, the month's result is 150,000 + 136,000 − 240,000 = +$46,000. Looks healthy.

Double the acquisition budget at the same CAC: 8,000 customers, $480,000 spend, $272,000 first-order contribution. The month becomes 150,000 + 272,000 − 480,000 = −$58,000. The P&L now reads as a loss, and a reviewer sees a business moving the wrong way, when in the cohort view the second month's customers are as valuable as the first cohort's and will pay back on the same schedule.

The monthly P&L measures cash conversion of the past. The cohort P&L measures return on the decision you're making today.

Building the cohort table

You don't need a data warehouse. You need an order-level table with customer ID, order date, order contribution, and the first-order date and channel for each customer. From there:

• Assign each customer to the month of their first order. That's their cohort for life.
• Divide time by cohort age, not calendar month. Compare every cohort at month 3, month 6, month 12, rather than January against July.
• Use contribution margin, not revenue. Revenue cohorts hide discount depth and return rates. The contribution margin guide covers the cost stack to include.
• Load acquisition cost per cohort, by channel where possible, to get cohort-level payback.

What to read from it

Three numbers do most of the work. First, payback month: when cumulative cohort contribution crosses cumulative acquisition spend. Second, 12-month and 24-month return: cumulative contribution divided by spend, so you can compare against what you'd earn on that capital elsewhere. Third, cohort-over-cohort drift: whether this quarter's cohorts are at the same place at the same age as last year's. If the January cohort had $41.60 per customer at month 3 and the June cohort has $37, you have an acquisition quality problem months before it appears in any blended report.

This is also where the lifetime value estimate gets checked against reality. A modeled LTV is a forecast; a cohort table is the actuals, and the gap between them is the first thing to review.

Turning payback into a spend decision

The cohort view turns "how much should we spend on acquisition" from a feeling into a constraint. You set the maximum payback period your cash can support, say 12 months, and spend at the level where cohorts meet it. In the example above, a 15-month payback fails a 12-month rule even though the 24-month return is 1.26x. You'd either lower CAC, lift early repeat orders, or fund the gap deliberately with a plan for the cash.

Segment the same table by channel before you set the ceiling. A blended 15-month payback can hide a paid-social cohort at 20 months and an organic or email-driven cohort at 6. The decision to cut or scale is rarely at the blended level; it's the channel mix inside it, and this is where the cohort table earns more than any dashboard of blended CAC.

Which brings in working capital: a brand adding a cohort each month carries a rolling negative balance until the older cohorts catch up, a cousin of the dynamics in the cash conversion cycle. Growth that is cohort-profitable can still run a company out of cash if nobody modeled the carrying cost.

See what your cohorts are actually doing

Most brands that build the table for the first time discover two things: their best channel and their best cohort are not what the blended report suggested, and payback is later than assumed. Both are easier to learn from a spreadsheet than from a bank balance.