Customer Clusters

Clusters helps with personalization and allow to reduce promotions without sacrificing revenue.


What is clustering

Clustering is a machine learning method that groups customers based on their shopping behaviour.

Why clustering

Clusters helps with personalization and allow to reduce promotions without sacrificing revenue.

What it brings me

It enables you to connect producers with customers and create unique, personalized offers.

Clustering of shopping missions

Identify key shopping missions

Using cluster analysis of shopping history

Customer segmentation

Assign customers to shopping missions

Update regularly for every customer

Product matching

Match promo products to shopping missions

Each shopping mission has linked products

Communicate personalized offers

Communicate personalized offers

Leaflet newsletter
Push notifications


Optimize offers

Utilize findings from shopping missions

Optimize promoted products and discounts

Personalized discounts

Introduce personalized discounts

Discounts only for selected customers


Detailed purchase numbers enable us to identify patterns in customers behaviour and cluster them accordingly.

Algorithm can generate any number of clusters, with 6-8 clusters often considered optimal from business perspective.


Before algorithm can do its work, we need to define predictive categories. For example, this 24 categories.
  • Beauty
  • Beers
  • Bread
  • Cheese
  • Cooking
  • Dairy
  • Deli
  • Durables
  • Frozen
  • Fruits
  • Ham
  • Healthy
  • Kids
  • Meat
  • Pet
  • Snacks
  • Soft Drinks
  • Spirits
  • Sweets
  • Tobacco
  • Tea & Coffee
  • Veggies
  • Wine
  • Yogurt

Clustering algorithm


Predefined number of most significant clusters is defined

Clusters are described as personas

Every customer belongs to 1 cluster

Affinity to all other clusters is calculated

Algorithm provides a way how to reassign customers to clusters on regular basis



After the several iterations the resulting clusters may look for example like this

Cooking cluster

  • Customers loving fresh products (fruit & veg)
  • Buys products for cooking & baking
  • Not buys alcohol and meat
  • Smallest segment

Wine cluster

  • Customers heavily buying wine
  • Average buyers of other alcohol
  • Not buying Yogurt, bread or dairy
  • Average sized segment