How can I test different product recommendation strategies?

Testing Product Recommendation Strategies

Product recommendations are one of the highest-AOV levers available in cosmetics e-commerce. But different recommendation strategies - algorithmic, manual, bestseller-based, routines-based - perform differently depending on store size, product catalog depth, and customer behavior. Testing different approaches is the only reliable way to identify what works for your specific audience.

Recommendation Strategy Comparison

Testing Framework

• Primary metric: AOV for orders that included a recommended product vs. those that didn't - this measures actual revenue impact, not just click-through
• Secondary metric: Recommendation click-through rate and add-to-cart rate from recommendations
• Test duration: 2-4 weeks minimum for cosmetics, which has natural weekly purchase cycles
• Traffic allocation: Split by session rather than by user to minimize contamination between test arms

Growth Suite Recommendation Integration

Growth Suite's Frequently Bought Together uses actual co-purchase data from your store to surface the most statistically validated product pairings. Advanced Cart Drawer surfaces these recommendations at the cart stage - the highest-receptivity moment for additions. Trending Products offers an alternative strategy for visitors who haven't made product choices yet, surfacing popularity-based social proof rather than purchase-based personalization. Testing these approaches against each other with the built-in A/B Testing Module provides the data to optimize recommendation strategy over time.