Machine Learning

Product Recommendations

The Product Recommendations model predicts a set of products likely to interest a specific user.

  • For a given user, this model recommends a set of products likely to interest the user.

  • For a given product, this model recommends a set of users who are most likely to be interested in the product.

CDP can apply post processing business rules to promote a specific type of products.

Use cases

Automation and personalization

Marketing communication is personalized and automated from the content personalization within CDP. The marketing team focuses on building creatives and the marketing strategy. The Acquia recommender system mines data to find the right product for each customer.

Timing of email campaigns

An individual’s recommendations remain the same unless the browse or purchase event occurs. Therefore, Acquia’s use cases are event-based:

  • Post purchase follow up (typically 14-30 days)

  • Alternative to Abandoned Browse (2-4 days)

  • Lapsed Customer (8 months inactivity)

How does the model work?

CDP’s recommender system is based on collaborative filtering techniques. The recommender system:

  • Is a frequency-based algorithm based on the purchase and browsing history of most users.

  • Mines the order history for highly correlated products.

  • Considers that the frequency of product pairs or groupings exceed a minimum confidence threshold.

  • Is biased toward recent purchases.

  • Removes outliers such as top 1% buyers and top 1% transactions.

  • Performs regular computations as needed. For example, a daily or weekly computation.

  • Excludes items recently purchased by the customer in personal recommendations.

Using the Product Recommendations model

CDP displays the output from the Product Recommendations model in:

  • 360 > 360 Profiles

  • Actions > Campaigns and Actions > Campaigns+

To learn more about how the Product Recommendations model can enhance your marketing workflow, contact us.