Cohort Analysis
This feature is currently available only on BETA.
Cohort Analysis groups your customers by when — or through which channel — they first bought from you, and then follows each group over time. It answers the question most reports can't: not "how did this channel perform last month", but "are the customers we're winning actually worth it, and are they coming back?"
When to use it
Reach for Cohort Analysis when you want to know:
Are newer customers more or less valuable than the ones we acquired six months ago?
Which campaign brings one-off buyers vs. loyal repeat customers?
Did a specific launch, price change, or promo produce a better or worse cohort than usual?
Do customers from one country, device, or product line retain better than others?
The 6 KPIs
The cards at the top give you the headline numbers for the customers you've selected. The date picker decides who counts: everyone whose first purchase falls in the chosen period. Those customers are then tracked all the way up to today.
Customer Acquisition Cost
What you paid, on average, to win one new customer.
Total spend ÷ new customers
Avg. Order Value
How much a typical order from these customers is worth.
Total revenue ÷ total orders
Avg. Time to Second Purchase
How quickly new customers come back for order number two.
Average days between 1st and 2nd order
Returning Customer Rate
The share of new customers who bought again at least once.
Customers with 2+ orders ÷ all customers
Avg. Number of Orders per Customer
How often an average customer buys from you.
Total orders ÷ customers
Avg. Lifetime Value (LTV)
The total revenue an average customer brings across all their orders.
Total revenue ÷ customers
How to read the report
The cohort table
Each row is a group of customers — one acquisition period (a week, a month…) or one value of the selected dimension (Platform Name, Campaign…). Each column shows how that group is doing as it ages: the first column is their starting point, and every column after it is one step later in their life as customers. Colour shading makes it easy to spot at a glance which groups grow and which fade. Hover any cell for the full detail.
The performance chart
The line chart below the table puts every cohort on the same starting line, so you can compare their shape regardless of when they started. Shows the first 5 cohorts by default; you can pick up to 10.
Cumulative vs. Absolute
A switch in the table header changes what each column means:
Cumulative — a running total. Good for "how much has this cohort brought in so far".
Absolute — just that single period. Good for spotting when activity happens or drops off.
Settings
Metric
What the table and chart measure — e.g. Net Revenue, Conversions.
Dimension
How customers are grouped. Date = group by when they were acquired; a channel dimension (e.g. Platform Name, Campaign) = group by where they came from.
Granularity
The size of each period: Daily, Weekly, Monthly, Quarterly or Yearly.
Number of periods
How far into each cohort's life you track it — i.e. how many columns you see.
How customers are counted
A "customer" is a unique person identified by their customer ID, with all their orders linked to them over time. Their first order makes them a new customer in a cohort; any later orders count as repeat. Importantly, a customer is not the same as an order — someone with 3 orders in January still counts as 1 customer in the January cohort.
How a customer is assigned to a cohort depends on the dimension:
Date dimension — one customer, one cohort. A customer belongs to exactly one cohort, decided by the date of their first order. All their future revenue and orders are tracked under that single cohort.
Other dimensions (e.g. Platform Name) — a customer can be split across cohorts. Because a customer's journey often touches several channels, they are shared fractionally across those channels according to the selected attribution model's credit split. How the credit is divided depends on which model you pick.
Filters
Filters respect where a customer started. A customer stays credited to the cohort and channel of their first purchase — so if you filter to a channel, you see the customers that channel acquired, including any revenue they later brought through other channels. (A handful of order-level filters, such as Conversion Type, Delivery Type or Order Status, apply to all orders instead.)
Exporting
Both the table (XLSX) and the chart (PNG) can be exported. All of them — plus KPIs — can be added to a board.
When is data available?
Cohort Analysis needs at least one site with a transaction feed. If your account doesn't have one yet, contact us and we'll help you set it up.
Last updated
Was this helpful?