Data used to train Algorithms

The data used to train AB Tasty Recommendation & Merchandising algorithm is provided through the integrations you set.


When there are multiple integration sources, the data can come from any of those integrations.
The recommendation algorithms, such as similar_items and associated_items, use different data analysis periods depending on the data sources:


Google Analytics 4 (GA4):

  • Transactions: 90 days.
  • Product views: 120 days.Data from GA4 is only available starting from the date the BigQuery GA4 connection was created.

Shopify:

  • Transactions: limited to 60 days.

PrestaShop:

  • Transactions: 90 days.

These periods are defined by the limitations of the data sources and ensure efficient analysis for recommendations.

Data used to train Recommendations Algorithms

Product pages viewed in session
sessionID + pageUrl (+ userID optionnally) for each pageview
integration source: Analytics

 

Transactions

transactionID + itemID (+ userID optionnally) for each transaction
integration source: Analytics, Transaction feed, CMS

 

Catalog

itemID + pageUrl + …itemProperties (+ userID optionnally) for each item
integration source: CMS, Catalog feed

 

User

userID + …userProperties
integration source: CRM (optional)

Was this article helpful?

/