A Recommendation Strategy is a reusable configuration that defines how personalized product recommendations are displayed in your marketing campaigns.
Instead of setting up recommendations from scratch for each Smart Recommender campaign, you can create a strategy once and apply it across multiple campaigns. This approach:
Ensures consistency in how recommendations are delivered.
Saves time by increasing efficiency.
Allows you to refine and update strategies centrally without modifying individual campaigns.
To manage all your strategies, navigate to Recommendation > Recommendation Strategies, where you can create new strategies or edit existing ones.
Why are Recommendation Strategies important?
Recommendation Strategies streamline your campaign management and improve the relevance of product recommendations. Instead of configuring rules separately for every campaign, strategies act as a central hub: you set them up once and reuse them across multiple channels and campaigns.
Recommendation Strategies help you:
Boost Conversions: Deliver personalized recommendations tailored to user behavior and preferences.
Enhance Engagement: Keep users interested by showing relevant products based on their browsing and purchase history.
Save Time with Centralized Control: Build strategies once and easily reuse them across campaigns, instead of recreating rules every time.
Optimize Performance at Scale: Update or fine-tune a strategy in one place and instantly apply changes everywhere it’s used.
Best Practices for Recommendation Strategies
Choose the right algorithm
Algorithm Selection: Visit the Smart Recommender Algorithms page to review all available options and choose the algorithm that best aligns with your campaign goals.
Algorithm Health: Each algorithm has specific prerequisites to deliver effective recommendations. To monitor this, go to the Recommendation Algorithms and Settings.
A green icon indicates that prerequisites are met and the algorithm is ready to run.
A red icon indicates that prerequisites are not met, meaning the algorithm may use a fallback or fail to generate recommendations.
Preview your strategy results
Before you publish or reuse a strategy, preview how it performs to ensure the results match your expectations. The Preview feature helps you validate filtering rules, confirm algorithm behavior, and check whether the strategy returns enough items for your campaigns.
When you open a strategy, use the Preview section to:
Check Recommended Items: See a sample of products your strategy will recommend based on the configured algorithm and filters.
Validate Filters: Confirm that filters such as price ranges, categories, stock status, and exclusions work as intended. If the preview returns too few items, adjust your filters or algorithm choice.
Evaluate Personalization: For strategies that rely on user behavior, select a test user to see how the recommendations change. This helps verify that personalized algorithms, such as User Based or Recently Viewed align with real user journeys.
Use the Preview feature to refine your strategy before using it in live campaigns.
Optimize recommendation results
Balance Filters and Exclusions: Avoid over-filtering, which may reduce the number of available recommendations.
Leverage Attribute Affinity: For highly relevant results, consider using the Attribute Affinity algorithm to tailor recommendations based on user preferences. Boost personalization by prioritizing user-preferred attributes like brand, color, or category.
Use Shuffling: Prevent static recommendation patterns by randomizing product order.
Configure Your Algorithms: Customize your algorithms' event sources and look-back periods to obtain the best results. To configure, navigate to the Recommendation Algorithms and click the Configure button on the top right.
Look-Back Period: The look-back period defines how many days of user event data an algorithm uses. Adjusting it helps capture trends or seasonality, such as setting 7 days for a “Hottest of the Week” campaign. You can increase the period (e.g., 14–30 days) during low-traffic times to improve accuracy. Different algorithms need different amounts of data, so choosing the correct look-back period is key for effective recommendations.
Data Sources: Different users interact with your brand in different ways, such as some visit your website, others use your app, or shop in-store. Each of these users has unique behaviors that the recommendation engine takes into account. You can control which data sources are used for each algorithm by selecting from the “Web,” “App,” and “Offline” options to ensure your recommendations are tailored to the specific behaviors of each audience, leading to more relevant results.
Troubleshoot common issues
Recommendations not appearing
Check if the assigned strategy has enough recommended products.
Verify that the filters and exclusions are not too restrictive.
Check your selected algorithm’s health from the Recommendation Algorithms page to make sure it’s healthy.
Make sure that item IDs in your user event data match the item IDs in your Product Catalog.
Strategy not updating across campaigns
Confirm that the strategy is active.
Refresh the campaign or wait a few minutes for updates to reflect.
Attribute Affinity not working
Ensure the Attribute Affinity feature is enabled in Smart Recommender settings.
Check the Product Attributes to ensure the attribute toggles are enabled for Attribute Affinity.
Check if the selected attributes align with the user event data and the Product Catalog.