Upload a Recommendations Dataset

This content is only for clients with Product Data and Customer Data options in the DATASETS menu in the top navigation bar. If no menu options appear when you click DATASETS, then this content doesn't apply to your account. Instead, see the alternate version of this documentation.

Follow these steps to create a Recommendations dataset.

Creating the Dataset Schema

You don't need to upload a full file when you create a new Recommendations dataset schema. A file that contains the required headers and at least three rows will suffice.

All the attributes required for a Recommendations dataset must appear in the file header, with the attribute name matching the exact spelling and case. See the dataset specifications in the Recommendations dataset overview to ensure success.

Click DATASETS in the top navigation bar, and then click Product Data.
Click CREATE DATASET.
Click Recommendation and then click NEXT.
Name the dataset and then click NEXT.
The name must begin with a letter and can contain only letters, numbers, and underscores.
Drag and drop the file into the upload area, and then wait for the platform to validate the data.
If the file fails validation, an error message appears that states the problem encountered. Revise the file, click RE-UPLOAD, and then drag and drop the revised file into the upload area.
After the file passes validation, ensure the sample shown is correct, select Looks good, and then click CONFIRM AND CREATE.

Once the platform finishes creating the schema, the details page for the newly created Recommendations dataset loads, with the Current Updates table showing the file you just uploaded listed and the PENDING status at the top-right corner.

The details page for a new Recommendations dataset, with a callout of the PENDING status

With the schema created, you can now update the dataset as needed, especially if you used a file that only contained a few rows when creating the schema. A full update replaces the entire dataset, and a partial update allows you to add new rows or update existing ones.

Updating a Recommendations Dataset

You can make updates using the Data API, SFTP, or the Upload Data wizard in the platform. Note that integrating with the API enables more real-time use cases, such as removing out-of-stock products from the dataset.

You cannot add attributes, also called fields or columns, to a Recommendations dataset. This dataset type can only contain the lookup_key, id, and rank attributes. The platform discards any other attributes and causes the update to fail. The File Upload Error modal notes the presence of the unknown field(s).

To successfully update the dataset, remove the disallowed field(s) and then upload the revised file.

When using the Upload Data wizard to update a Recommendations dataset, you can either drag and drop a file directly into the wizard, or you can input a URL to a dataset file that you store on Amazon S3. Regardless of which of these two upload options you select, the file that you use must be in the CSV or TSV format.

Uploading a File Directly

Follow these steps to update a Recommendations dataset by uploading a dataset file directly to the Monetate platform using the Upload Data wizard.

Click the upload icon that appears to the far right of the dataset's entry on the All Datasets list page or the Recommendations list page, and then select Manual File Upload.
Select the type of update you want to perform, and then click NEXT.
Select Manual File Upload and then click NEXT.
Drag and drop the file into the wizard.

Monetate ingests the file and closes the Upload Data wizard. When the dataset's Update History tab loads, a message briefly appears that informs you that the file was successfully uploaded. Additionally, PENDING appears where the timeframe of the last update normally appears.

Callout of the upload success message and the PENDING status on the Recommendations dataset's details page

If the file failed to upload or the platform encountered other errors, you see messages indicating the problem. See Failed Updates in this documentation for more information.

Uploading a File Using an Amazon S3 URL

Follow these steps to add rows to or update existing rows in a product catalog by uploading a dataset file using an Amazon S3 pre-signed URL.

Generate the pre-signed URL for the file you want to upload.
Refer to Sharing objects using presigned URLs in the Amazon Simple Storage Service (S3) user documentation for more information about generating a URL for this purpose.
Click the upload icon that appears to the far right of the dataset's entry on the All Datasets list page or the Recommendations list page, and then select Manual File Upload.
Select the type of update you want to perform, and then click NEXT.
Select Upload File Using Pre-signed S3 URL, and then click NEXT.
Paste the Amazon S3 URL into the field, and then click FINISH.

The Upload Data wizard closes and Monetate ingests the file. When the dataset's details page loads, a message briefly appears that informs you that the file was successfully uploaded from the Amazon S3 URL. Additionally, PENDING appears where the timeframe of the last update normally appears.

A Recommendations dataset's details page, with a callout of the file upload success message and of the 'PENDING' status

If the file failed to upload or the platform encountered other errors, you see messages indicating the problem. See Failed Updates in this documentation for more information.

Failed Updates

If the platform encounters an error with the update, you see UPDATE FAILED in multiple places within the platform:

In the LAST UPDATED column for the dataset's entry on the All Datasets list page and the Recommendations list page
In place of the last update's timeframe on the dataset's details page
In the UPLOAD DATE column in the Update History table