Recommendations Datasets Overview

In addition to the recommendation algorithms and filters that you use to build recommendation strategies and bundles for Monetate Dynamic Bundles, you can upload your own product recommendations as a Recommendations dataset. See Create a Recommendations Dataset to learn how to accomplish that task.

Dataset Specifications

This table contains the attributes, also called columns or fields, you must include in a Recommendations dataset.

AttributeData TypeExampleDescription
lookup_keyStringabcValue used to find related recommendations. These values should associate to item_group_id parent identifier values in the linked product catalog.
idStringxyz-123A product's unique identifier. Use the SKU collected in your web integration where possible. These values should associate to the id values in the linked product catalog.
rankNumber10The ranking of importance or the position in which this record is returned in the list of records from the specified lookup_key.

Only the attributes in the specifications table can appear in a Recommendations dataset. If you include additional ones when creating the dataset schema, Monetate discards them without alerting you. If you include them when updating a Recommendations dataset, the update then fails, with the File Upload Error modal noting the presence of the unknown field(s).

The 'File Upload Error' modal with information about an unknown field in a Recommendations dataset file

Submit a support ticket using the Monetate Technical Support portal ( if you need additional assistance customizing recommendations.

Use Cases

Here are three examples of how you can use a Recommendations dataset.

To leverage a Recommendations dataset, ensure the linked product catalog contains the values defined for item_group_id and id, respectively.

Display Customized Recommendations

Purpose: To create your own product recommendations for any products

Implementation: Input an item_group_id (product ID) in the lookup_key column, and list in the id column a product that you want to recommend for the corresponding item_group_id.


  • Leverage the expertise of merchandisers to offer curated recommendations for specific products.
  • Leverage the output of a new algorithm internally developed by a data science team.

Sample file: RecDataset.csv

Recommend a Specific List of Products

Purpose: To display a manually curated list of products

Implementation: Ensure the lookup_key column is populated with a value of none for all line items.


  • Display a curated list of products on a homepage or landing page.
  • Display products that you want to highlight as "New Arrivals."

Sample file: RecDataset_none_lookupkey.csv

Recommend Products to Specific Customers

Purpose: To display internally developed recommendations for individual users

Implementation: Ensure that the lookup_key column is populated with customer IDs and that the id column is populated with the products that you want to recommend for the corresponding customer IDs.


Leverage the output from a data science team to present individually curated recommendations to high-value shoppers on a landing page.

Sample file: RecDataset_custid_lookupkey.csv