Recommendation feeds import from data warehouses

You can start your feed import by navigating to the import recommendations overview page and selecting the appropriate import type.

You can import data from a data warehouse using one of the following methods:

  • Databricks – Connect to Databricks and import data from a specified database and table.
  • Snowflake – Import data from Snowflake by providing account credentials, database, and schema.
  • Google BigQuery – Fetch data from BigQuery using service account credentials and specifying the dataset and table.

Choose the appropriate method based on your data source and access preferences.

Left-side panel overview

  • Source: Displays the data source for the import. Example: Google BigQuery.
  • Destination: Specifies where the imported data will be directed, i.e., recommendation feeds in this case.
  • Select recommendation type: (mandatory field)
    • User-to-product: Imports recommendations based on user identifiers and related product IDs, ideal for personalized promotions.
    • Product-to-product: Suggests related products based on user interactions, such as viewed or purchased items.
  • Recommendation feed name: (mandatory field) Enter a name for the feed (minimum of 3 characters).
  • Description: Add details to describe the recommendation feed.
  • Notification preferences:
    • To: Specify email IDs to receive notifications about the import task.
    • CC: Add additional recipients to be copied on notifications.
    • Settings:
      • Notify - any status: Sends notifications for all import task statuses.
      • Notify - If more than [X]% of records fail: Sends alerts when failures exceed the defined threshold (e.g., 70%).
Recommendation-Import-Left-Panel.png

Source configuration

To configure the data source for import:

  • Select an adapter – Choose the appropriate adapter for your import.
  • Import from – Choose between:
    • Table – Selecting this option reveals a Table dropdown where you can choose the specific table to import.
    • View – Selecting this option reveals a View dropdown where you can choose the appropriate database view.
  • Click Continue to proceed.

Data configuration & validation

  • Sample Data: Displays 10 records fetched from the source to assist in field mapping.
  • Field Mapping:
    • Map the Source Attribute Name to the corresponding Destination Attribute Name and specify the Destination Data Type.
    • Mapping Instructions:
      • Only columns mapped to a destination attribute will be imported.
      • Map Floating point numeric data types from the source to Decimal in Blueshift.
      • Example: For events like purchase, add_to_cart, wishlist, or view, map a column to product_ids (required for events interacting with product catalogs).
      • Ensure one column is mapped to a customer identifier, such as customer_id, email, my_custom_id, cookie, device_id, or my_external_id.
Recommendation-Import-Configuration.png

Data Quality Check:

  • Use Check Data Quality to validate field mappings.
  • View data quality percentages, errors, and hints for corrections in the modal.
  • Adjust mappings as needed.
Recommendation-Import-Data-Quality.png

Test Run:

  • Click Test Run to validate with up to 10 records.
  • The modal displays source data alongside mapped JSON.
  • Update mappings and re-run tests if necessary.
Recommendation-Import-Test-Run.png
  • Next Steps: Address all validation issues and click Continue to proceed with the import.

Additional configuration

  • Type of import:
    • Full import – Imports the entire table or view for each iteration.
    • Incremental import – Imports only new or updated data for each iteration.
  • Difference identification (for incremental import):
    • Select Time or Number Based to identify incremental records using an attribute from the source data.
  • Difference identifier (for incremental import):
    • Choose the attribute used to track incremental changes (e.g., joined_at).

Click Continue to proceed.

Scheduling and launching the import task

  • Select the Start Date using the date picker.
  • Check 'Is it a recurring data import?' to enable recurring imports.
  • Choose when the task ends:
    • 'Never' for an indefinite schedule.
    • 'At some time' to set an End Date.
  • Set the execution frequency (e.g., every 15 minutes).
    • Scheduling options: Minutes, Hourly, Daily, Weekly, and Monthly.

Review the setup and the top right corner of the screen:

  • Click the Save button to save the task.
  • Click the Launch button to start the task.

Recommendation-Import-Launch.png

Import errors and possible causes

Some records may not be imported due to errors. Below are common errors and their possible causes:

Error message Possible cause
Firefly resource not found error={"{\"errors\": [\"attribute_value_not_found\"]}"} An item in the product_ids list is not present in the catalog.
Firefly resource not found error={"{\"errors\": [\"product_id_not_found\"]}"} The source_product_id itself is not in the catalog.
Was this article helpful?
0 out of 0 found this helpful

Comments

0 comments

Please sign in to leave a comment.