A/B Testing
What You Can Test
Configurable settings for each experiment variant
Each variant in an A/B test can be configured independently across all prevention settings. This lets you isolate exactly which change drives better results.
Testable Settings
| Setting | What You Can Compare |
|---|---|
| AI Prevention toggle | Enabled vs. disabled — the classic holdout test |
| Incentive type | Percentage discount vs. fixed amount |
| Incentive amount | e.g., 10% off vs. 15% off |
| Minimum purchase threshold | No minimum vs. a specific cart value |
| Discount stacking | Whether the NavonaAI discount combines with order, product, or shipping discounts |
| Product/collection exclusions | Restrict discounts to certain parts of your catalog |
Common Experiment Scenarios
Holdout Test (Recommended First Experiment)
The most valuable experiment to run first:
- Control: AI prevention OFF (no popup shown)
- Treatment: AI prevention ON with your current settings
This tells you definitively how much lift NavonaAI is generating for your store.
Discount Amount Optimization
- Control: 10% discount
- Treatment: 15% discount
Find the sweet spot between conversion rate and margin impact.
Discount Type Comparison
- Control: 10% percentage discount
- Treatment: €5 fixed amount discount
Test which format resonates better with your customers.
Minimum Purchase Threshold
- Control: No minimum purchase required
- Treatment: €30 minimum cart value
Determine if a threshold protects margins without hurting conversion.
For the clearest results, change only one variable between your control and treatment. If you change the incentive type and the amount at the same time, you won't know which change drove the difference.