35. More Real-World Project Briefs

35.8 Customer Analytics and a Simple Cohort

By Ravindra Bagale · Last updated: 6 October 2026

Business question. Who bought in more than one month, and which join-month cohort still bought in a later month?

Tables. Order (Order ID, Date, Customer, City, Amount). Customer (Customer, City, First Order Date). Use only the class names. Date.

Measures.

Customers = DISTINCTCOUNT(Order[Customer])
Repeat Customers =
COUNTROWS(FILTER(VALUES(Order[Customer]), [Orders] > 1))
Repeat % = DIVIDE([Repeat Customers], [Customers])

[Orders] is DISTINCTCOUNT(Order[Order ID]) in this brief.

Cohort, kept small. A cohort is the set of customers who share a start month. Add Cohort Month on Customer from First Order Date (Power Query, start of month). A matrix with Cohort Month on rows and order month on columns, and [Customers] in values, shows how many of that start group ordered later. It will be sparse on ten people. That is fine. Do not smooth it.

Pages. Cards: Customers, Repeat %. A matrix for the cohort. A bar of sales by customer. A city slicer.

Sample layout — fictional data. The picture uses the worked rows in this brief. It does not add a table.

Fictional read: Ruhi Bagale's first order is in January in Pune, and she orders again in March. She is in the January cohort and appears in the March column. Amir orders only in February in Nagpur. He is a February cohort with one column. If Ruhi appears in every cohort, First Order Date is calculated with a measure that still sees the visual filter. Compute First Order Date in Power Query so it does not move.

Worked rows (fictional sample data). First order month is a column, not a measure (see the warning above).

Distinct customers are Ruhi, Zoya, Rani, Amir, Salman, Raja. That is 6. Repeat customers are people with more than one order: Ruhi and Zoya. That is 2. Repeat % is 2 ÷ 6 = 33.3%.

January cohort is Ruhi, Zoya, and Rani (3). In the March column of that cohort only Ruhi appears (1). Amir is a February cohort with one order. If Ruhi also appears in a February cohort, First Order Month was calculated under a visual filter. It must not move.

Starter file: brief_35_8_customer_orders.csv – the worked rows above as a CSV (fictional practice data). Load it with Get data › Text/CSV, build the measures, then check your cards.

What the manager does. The Pune manager asks what Zoya bought in February, because she is a repeat, and does not add a "win-back" campaign for Ruhi in the same breath without reading the row. The Nashik manager sees Rani once in January. She is not a repeat. A Nashik card that says "orders exist" is the vanity metric. The action is a second-order offer for Rani, or an honest blank repeat card.