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). Implemented with tf–idf.
|Model Update Interval||3 days|
|Can Manually Update Model||No|
|Can Preview Output||No|
|Recommendations Build Time||This model will take some time to build. Depending on the number of products your store have, it may take anywhere from a few minutes to hours.|
- Products must be at least 10% similar.
Where should you place this recommendation?
Since this recommendation model doesn't require any past purchase data, you can use it as fallback for newly added products that wouldn't have any "Frequently Bought Together" and "Also Bought" recommendations. See guide on using fallback recommenders.