The Product Clustering model is a statistical analysis tool that groups customers by their product category preferences. 

Product Clustering models analyze data from Customer Data Platform (CDP) customers who have made more than one purchase in the past three years. The results are used to identify the patterns in buying behavior. The model creates clusters of customers with similar buying interests.

After a model has been trained using your CDP data, you can use the results by choosing customer clusters that align with your campaign goals. Using the campaign creation tool, you can use the clusters to build targeted lists for specific audiences.

Using the results of the Product Clustering model

1. Sign in to the CDP user interface.
2. Navigate to Actions > Campaigns+.
3. Click New Campaign.
4. On the Setup page, enter a name and description for your campaign.
5. Click Next.
6. On the Audience page, in the Include customers who area, click Add Group.The system displays the pop-up window.

The Product Clustering model is a statistical analysis tool that groups customers by their product category preferences. 

Product Clustering models analyze data from Customer Data Platform (CDP) customers who have made more than one purchase in the past three years. The results are used to identify the patterns in buying behavior. The model creates clusters of customers with similar buying interests.

After a model has been trained using your CDP data, you can use the results by choosing customer clusters that align with your campaign goals. Using the campaign creation tool, you can use the clusters to build targeted lists for specific audiences.

Using the results of the Product Clustering model

1. Sign in to the CDP user interface.
2. Navigate to Actions > Campaigns+.
3. Click New Campaign.
4. On the Setup page, enter a name and description for your campaign.
5. Click Next.
6. On the Audience page, in the Include customers who area, click Add Group.
   The system displays the Add Rule pop-up window.

7. Select Machine Learning Segments.

   
8. Select a product cluster segment.
9. On the Audience page, use the cluster names to reference one or more clusters.
   1. For one cluster, use Equal to.
   2. For multiple clusters, use Contains.

10. Click Next and complete the campaign creation process.

    For detailed steps, see Creating Campaigns.

Naming clusters

To simplify the process of using product cluster segments, Acquia supports implementation of custom segment names. You can name the clusters to match your business requirements so that the users can easily identify the correct clusters for their campaigns. The cluster names are visible wherever the model operates, for example, within the Campaigns+ interface. During the build, each cluster is named based on the most relevant product range. You can either retain the names or customize them to match your requirements.

Training results visual example

The pie chart shows your customer base divided into seven distinct groups. The percentage of each group represents the portion of the total qualified customer base of an apparel company. 

The following example shows how to name the clusters based on the most relevant item in each product list:

• Cluster 0: Executive Attire
• Cluster 1: Leisure Apparel
• Cluster 2: Athletic Gear
• Cluster 3: Seasonal Outerwear
• Cluster 4: Premium Footwear
• Cluster 5: Fashion Accessories
• Cluster 6: Home Leisurewear

Insights from cluster-based ML model output

Cluster 0: Executive Attire

Summary: A graph of this shaping shows that multiple product lines are relevant to the cluster of customers. Cluster 0 indicates that these customers prefer executive attire for example, having already purchased some items from that category. Marketers may want to feature more items in a customer's primary cluster or offer discounts to merchandise in alternative categories to encourage customers to increase cross-category spending with the brand.  

The following is an example of a cluster DNA, which illustrates the categories commonly purchased within that cluster and their ratios compared to the average customer.

Similarly, you can analyze other clusters to understand customer preferences.