This article explains how the add-on "AI similar products" works in Fynode, what data it uses, and how to configure it correctly for better cross-sell and up-sell recommendations. We describe the order analysis principle, available settings, and recommended steps for testing and deployment.
Principle of operation
The plugin analyzes orders for approximately the last half year and determines which products are bought together, how often they appear in the same order, and whether they share categories. Based on this analysis the system will propose a list of the most likely similar products customers add for each product. Recommendation updates run automatically, typically at night, resulting in setting related or up-sell items in the eshop.
Testing and deployment
After activating the add-on the system will display a random product and generated recommendations for quick testing. Use the "Generate again" button to see results for other products and tune settings based on recommendation quality. Before global deployment we recommend testing on one or several categories to check recommendation relevance and impact on UX. If everything is fine, activate repetition according to your preference and monitor results.
Key plugin settings
In settings, specify the “Relation” for how recommendations should be saved (Related, Up‑sell, Cross‑sell), the frequency of update repetition, and any restriction to specific categories. You can also enable the “Shared categories” logic so the system prefers products from the same categories and turn on “Require joint purchase” if you want to generate recommendations only based on actual joint purchases. You also set the “Maximum limit” of recommended products per product.
Costs and planning
In the plugin settings you will find informative pricing: price for processing a single product and an estimate of total costs based on number of products and chosen frequency. When recurring scheduling is enabled, the corresponding amount will be deducted from the AI credit at the set intervals. Always check the estimate before activation so you can control credit consumption.
What data does the plugin use to generate recommendations?
Historical orders from approximately the last half year are primarily used so the system can evaluate common purchases and co-purchase frequency. If these data are missing in the system, recommendations will be based more on internal product similarities than on real orders.
Can I restrict generation to only certain categories?
Yes. In the plugin settings you select specific categories for the AI to focus on. This is useful for testing or if you want to initialize the function first only for selected parts of the catalog.
How often are recommendations updated and how are they charged?
Set the update frequency in the “Repeat” field, for example every 7, 14, or 30 days. For each repeat an AI credit fee will be deducted according to the number of products and the set price, so check the calculator in the plugin settings to estimate total costs.
Hello, we are currently online, we will be happy to answer your questions.