Skip to main content
Five cohort reports follow what each cohort spends and orders over time. New to cohorts? Start with What is cohort analysis?

Read any cohort table in 30 seconds

Every report under Cohorts is laid out the same way, so once you can read one, you can read them all.

A row is a group of customers

Everyone whose first order fell in that month (or week, or year). The row’s label is when they joined, not when they ordered.

A column is time since joining

Month 1 is the month they joined, Month 2 the month after, and so on. It’s counted per customer group, not on the calendar.

Customers is the group's size

How many new customers joined that period. It never changes as time passes, so you can compare rows fairly.

Weighted Average is your benchmark

The top row combines every cohort, giving bigger cohorts more weight. Compare a row against it to spot a good or bad month.
Read columns, not just rows. Going down a column compares cohorts at the same age: “how did January’s customers do in their third month, compared with March’s in theirs?” That’s a fair comparison. Comparing cells on the same calendar month isn’t, because the cohorts are different ages.
A few things are the same on every report:
  • — (Period not yet reached) means the cohort isn’t that old yet, which gives the table its diagonal edge. It isn’t zero.
  • The last filled cell in each row is a period still in progress, so it’s usually low. Judge cohorts on completed periods.
  • Click a cohort’s name to open the Customers list filtered to exactly those customers, for example to email them or check where they came from.
  • Who’s counted: customers whose first counted order falls in your date range. Guest orders are grouped by email, and a customer refunded in full (nothing spent after refunds) isn’t counted. Clicking a cohort opens exactly the customers it counts.
  • Refunds come off in the period the refund was issued, not the period of the original order.

Lifetime Value

Lifetime value cohort table: cumulative spend per customer rising month by month
Cohorts → LifeTime Value answers: on average, how much has a customer from this month spent with you so far? Each cell is the cohort’s total sales to date, after refunds, divided by the number of customers in it. It’s cumulative: Month 6 includes everything spent in Months 1 to 5 too, so values only grow along a row (unless refunds outweigh new sales that month).
100 customers first ordered in March. Together they spent $5,000 in March, $1,200 in April and $800 in May.Month 1 is $5,000 ÷ 100. Month 2 adds April: $6,200 ÷ 100. Month 3 adds May: $7,000 ÷ 100. The divisor stays 100 even though most of them didn’t come back: it’s value per customer acquired, which is what you need for acquisition budgets.
The cards show the weighted average lifetime value at Month 3, 6, 12 and 24 across all cohorts old enough to have reached it. Not enough history yet means no cohort in your range is that old.
How much can I spend to get a customer? Take the Month 12 card and multiply by your profit margin. If customers reach $120 in a year at a 40% margin, each one earns you about $48, so paying $30 to acquire one is profitable; paying $60 isn’t.
Lifetime Value always groups by month and goes up to 24 months; the other reports also let you group by week or year.
What to look for
  • How steep the row is after Month 1. A big jump from Month 1 to Month 3 means repeat purchases make up much of a customer’s value. A flat row means most customers buy once.
  • Newer cohorts lower down a column than older ones at the same age: recent customers are worth less. Check what changed (discounts, a new ad channel, prices).

Average Order Value

Average order value cohort table
Cohorts → Average Order Value answers: when this cohort orders, how big are their orders? Each cell is the cohort’s sales in that period, after refunds, divided by the number of orders they placed in that period. It isn’t cumulative, and it only looks at customers who actually ordered that period.
**A $0 cell means nobody in the cohort ordered that period**, not that they placed $0 orders. With small cohorts, one or two orders can swing a cell a lot, so lean on the Weighted Average row and on cohorts with plenty of customers.
What to look for
  • Month 1 against later months. Month 1 includes the first order, often made with a welcome discount. If later months are higher, returning customers trust you with bigger baskets. If they’re lower, they may be coming back only for small top-ups or sale items.
  • Read it with Lifetime Value. A high AOV with a flat Lifetime Value means people buy big once and leave; a modest AOV with a rising Lifetime Value means a healthy repeat habit.

Orders per Customer

Orders per Customer cohort report: average orders per customer by month
Cohorts → Order Per Customer answers: in each period, how many orders did this cohort place for every customer in it? Each cell is the cohort’s orders in that period divided by the cohort’s full size (everyone, not just those who ordered). It isn’t cumulative.
100 customers joined in March. In March they placed 108 orders; in April, 15.Month 1 is 1.08: everyone placed their first order, and a few ordered twice that month. Month 2 is 0.15: roughly 15 orders for every 100 customers.Month 1 is always at least 1.00, because everyone in the cohort ordered that month.
Read a later-month value as “orders per 100 customers”: 0.15 ≈ 15 orders for every 100 customers acquired. Adding up a row gives the average number of orders a customer places over that time.
What to look for
  • Month 1 well above 1.00 means customers often come back within their first month, a sign of a strong first experience or a fast-consumed product.
  • Later months that hold steady rather than sliding toward zero point to a buying habit, which is what subscriptions, replenishment emails and loyalty schemes aim for.

Orders Over Time

Cohorts → Orders Over Time answers: how many orders did this cohort place in each period? Each cell is a plain count of the cohort’s orders in that period, not divided by anything, so big cohorts show big numbers.
The cells are totals, but the Weighted Average row and the cards are per customer. A cell of 42 means 42 orders; a Weighted Average of 0.18 means 0.18 orders per customer. To compare cohorts of different sizes, use Orders per Customer instead.
What to look for
  • How much work old customers do. Pick a recent calendar month and add up the cells for it across rows (they fall on a diagonal). That shows how many of that month’s orders came from customers acquired months or years ago.

Sales Over Time

Sales Over Time cohort report: sales by month for each cohort
Cohorts → Sales Over Time answers: how much did this cohort spend in each period? Each cell is the cohort’s total sales in that period, after refunds. Like Orders Over Time, it’s a total, not cumulative and not per customer. The Weighted Average row and the Avg Revenue per Customer cards divide by customers.
Sales Over Time is the per-period version of Lifetime Value. A row’s cells added up from Month 1, then divided by Customers, give that cohort’s Lifetime Value.
What to look for
  • Revenue you can count on. If older cohorts keep producing steady sales month after month, that’s recurring revenue you’d lose if you stopped serving them, which makes a good case for spending on retention.
  • A cohort that drops off a cliff after Month 1 usually came in through a one-off promotion. Check whether the campaign that brought them was worth it.

Which report should I use?

Using value cohorts

  • Budget acquisition with Lifetime Value at Month 6 or 12, not first-order value.
  • Spot a weak cohort early: if a cohort’s Month 2 is well below others, look at what was different when it joined (a deep discount, a new channel).
  • Holiday cohorts often spend a lot at first and return less; compare their Month 6 value with other months before repeating the campaign.
  • Group by Week for fast-repeat products (coffee, pet food) to see behaviour in the first weeks; group by Year for a long-term view of slow-repeat products.