List of Product Recommendation Types
Below are available product recommendation types you can use
Frequently Bought Together
Use frequently bought together to upsell commonly bought together products. Implemented with association rules.
Also known as "Customers Who Bought This Also Bought". Use this to cross-sell other products bought by customers. This algorithm looks at purchase history of customers who've purchased a given product. Then recommend other products based on their purchase history. Results will be similar to "Frequently Bought Together" if most of your customers are one-time purchasers.
Recommend best selling products by revenue. Use lookback days to get best selling products within a time period. Or filter by tags to get best selling products from a category.
Display recently viewed products based on browsing history.
Manually select products for recommendations. You can use this for various use cases like "Featured Items", "Recommended for Christmas", "Editors Favorites", "Staff Picks Under $20".
Recommend similar products based on textual attributes like title, description, keywords. The pros of this algorithm is that it does not require any past purchase data (aka cold start problem). Use this as fallback for newly added products that doesn't have sufficient order history to generate recommendations with other logics. Implemented with tf–idf.
Recommend trending products. Uses number of units sold to determine trending products. More recent sales will hold greater weightage.
Recommend products that are selling fast based on the number of units sold. Unlike trending, selling fast allows you to display the number of product units sold during period. Use trending if you don't wish to display this stat publicly.
Selling Out Soon
Recommend products that are low in stock. Display selling out soon to induce FOMO and get users to make quick purchase.
Recommend newly arrived products. Place new arrivals in your home page to drive that initial boost for new products.
Recommend discounted items. Use discounted items to capture user attention and increase the time they spent at your store.
Recommend top rated products. Add filter to exclude products with low number of reviews to make recommendations more compelling. The algorithm used is bayesian average, similar to the one used by imdb.com where a 4.8 rated product with 100 reviews would be ranked higher than a 5.0 rated product with 1 review.
Recommend recently sold products. Use recently sold products as an alternative for trending products.
Recommend recently reviewed products. Use recently reviewed products as an alternative for trending products.