Log in with GoogleMarket basket analysis
Market basket analysis finds which products are bought together by counting how many orders contain each combination of products.
By Berend Vrakking, founder of Product Metrics. Updated 7 October 2026.
Formula
Support of a pair = orders containing both products ÷ all orders
Data mining adds two more measures. Confidence is orders with both ÷ orders with the first product. Lift is that confidence ÷ the share of all orders containing the second product.
Example
| Count | Value |
|---|---|
| All orders | 100 |
| Orders containing Trail Runner 2 | 40 |
| Orders containing Running Socks | 50 |
| Orders containing both | 30 |
| Support of the pair (30 ÷ 100) | 30% |
| Orders with Trail Runner 2 that also contain Running Socks (30 ÷ 40) | 75% |
Three in ten orders contain both products, and three in four orders with Trail Runner 2 also contain Running Socks. Illustrative data.
For one product, and for an account
The useful question is per product: which other products turn up in the same orders, and how often. The account version, how many items an average order holds, is easier to report and says nothing about which product pulls the others along.
Product Journeys has a Purchase Basket Analysis view for each product, listing what else is in the basket when it sells and the share of baskets each product appears in. The clicked product is the entry product, so you can read the basket against the ad that started the journey.
Common mistake
Declaring two products a pair when one of them is in almost every order anyway. Everything ‘goes with’ the bestseller. Lift corrects for that by comparing with how often the second product sells on its own.
Questions
What is meant by market basket analysis?
What does lift indicate in market basket analysis?
Keep reading
- Cross-sellingWhat a customer buys beyond the product that was clicked.
- Product JourneysSee what else is in the basket when each product sells.
- ROASThe return figure for a product’s ad spend.