Basket Analysis Methodology
Key Metrics
Support = Transactions with Item(s) ÷ Total Transactions
How frequently items appear. "Bread appears in 30% of transactions" = 0.30 support.
Confidence = Transactions with A+B ÷ Transactions with A
If customer buys A, probability they also buy B. "Of customers who buy chips, 40% also buy salsa" = 0.40 confidence.
Lift = Confidence ÷ Support of B
How much more likely items are purchased together than by chance. Lift > 1 indicates positive association.
Example Analysis
| Association Rule | Support | Confidence | Lift |
|---|---|---|---|
| Pasta → Marinara Sauce | 8% | 65% | 4.2 |
| Ground Beef → Taco Shells | 5% | 42% | 3.8 |
| Wine → Cheese | 3% | 38% | 2.9 |
| Bread → Milk | 12% | 25% | 1.1 |
Pasta → Sauce has high lift (strong affinity); Bread → Milk has low lift (both common, weak association).
Actionable Lift
Focus on rules with lift > 1.5 and reasonable support. Very high lift with tiny support (rare items) may not be actionable. Very high support with low lift (common items) isn't insightful. The sweet spot is meaningful associations with sufficient volume.
Loyalty Program Applications
Personalized Recommendations
"Customers who bought X also bought Y" powered by basket data. Personalized suggestions based on current cart or purchase history.
Targeted Bonus Points
Offer bonus points on complementary products. If member bought pasta, offer 3x points on marinara sauce. Drives basket completion and higher AOV.
Category Expansion
Identify products that bridge categories. Use basket affinities to introduce members to new departments they're likely to enjoy based on current purchases.
Offer Bundling
Create promotional bundles based on natural affinities. "Complete the meal" offers combining frequently co-purchased items at bonus point rates.
Trip Type Analysis
Identify different shopping missions: quick trips, stock-up trips, occasion shopping. Tailor offers to mission type rather than treating all visits equally.
Segment Discovery
Basket composition reveals customer segments: health-conscious, convenience-focused, budget-minded. Use basket patterns for behavioral segmentation.
Basket Evolution Over Time
With loyalty data, track how member baskets change: new categories tried, brand preferences shifting, basket size growing or shrinking. This longitudinal view reveals member development and potential churn signals.
Best Practices
- 1. Set appropriate thresholds. Minimum support prevents rare coincidences from appearing significant. Start with 1-5% support threshold depending on transaction volume.
- 2. Look beyond obvious associations. Pasta-and-sauce is expected. Look for non-obvious affinities that reveal actionable insights—products from different categories with unexpected lift.
- 3. Segment the analysis. Basket patterns differ by customer segment, store format, day of week, and season. Run segment-specific analysis rather than one-size-fits-all.
- 4. Update regularly. Product assortments change, customer preferences shift, and seasonal patterns vary. Refresh basket analysis quarterly or when assortment changes significantly.
- 5. Test before scaling. High-lift associations suggest opportunity, but test offers on small segments before broad deployment. Correlation in baskets doesn't guarantee response to promotions.
- 6. Combine with other data. Basket analysis + member demographics + purchase history = powerful targeting. Use basket insights alongside RFM, CLV, and preference data.
Basket Analysis + AI
Modern offer decisioning systems use basket analysis as one input among many. Machine learning models incorporate basket affinities alongside member preferences, inventory levels, and business objectives to select optimal recommendations in real-time.
Exchange Solutions Basket Intelligence
Exchange Solutions' platform includes basket analysis capabilities—automated affinity discovery, segment-level basket patterns, and integration with personalized offer delivery. Turn basket insights into targeted loyalty offers that drive basket size and category expansion.
