Return quantity × product price, by return month. Own scale.
By country
Top 10 subcategories
Customer segments (RFM)
Revenue concentration (Pareto)
Customers sorted by revenue, cumulative share. The marker is the 80% line.
New vs returning revenue
Returning customersNew customers (first order that month)
Cohort retention by quarter
Row = customers whose first order fell in that quarter; cell = % of them ordering again N quarters later.
Basket network: what sells together
Node = subcategory, sized by the orders that contain it; link = a pair bought together more than chance (line weight = lift; pairs with ≥50 shared orders and lift ≥ 1.2). Drag nodes; hover for numbers.
Return rate by subcategory
Basket pairs by lift
Revenue forecast: next six months
ActualForecast ±95% band
Churn risk: repeat customers
Customer value by segment
Next best offer: from basket rules
Directional confidence: of orders containing the left item, the share that also contains the offer. Rules with lift ≥ 1.5 and ≥100 shared orders.
What-if levers
Static levers computed from the figures above.
Method caveat. Data ends 30 Jun 2022, so the forecast demonstrates the method on this dataset rather than predicting a current future. Churn tiers are rule-based (recency ÷ the customer's own median repurchase interval, floor 14 days), not a fitted model.
Cost = order quantity × product cost. Profit = revenue − cost (AR subtracts returns from revenue only; return-cost credit is open decision U-05).
Orders = distinct order numbers. AOV = revenue ÷ orders. Country = the order's territory (the Power BI model routes it via the customer, open decision U-03).
RFM: recency = days since last order at the scope's last order date; frequency = distinct (order, date); monetary = revenue. Quintiles 1–5, class = R·F·M digits, segments from Dim_RFM.xlsx.
Page 1 is computed live for the filters. Pages 2–3 are pre-computed per scope at each data load. Definitions match the semantic model (z_measure, z_RFM_analysis, z_Basket_Analysis).