Behavioural Segmentation

Behavioural customer segmentation aims to create different behaviour-based clusters of the customer base. Behavioural segmentation has more added features than RFM segmentation and it is not limited only to features like Recency, Frequency and Monetary. The outcome improves according to the features/context you input in this type of segmentation.

Behavioural segmentation allows you to do the following.

  • Understand the attitudes, likes, and dislikes of prospects and customers.
  • Identify customers who are most likely to buy from the org.
  • See what and when customers are most likely to buy.
  • Create more targeted content, messages, and marketing campaigns.
  • Monitor changes and growth patterns to develop predictive marketing plans.
  • Identify purchasing trends and uncover niche marketing.

Behavioural Features

Behavioural segmentation divides customers into groups according to their observed behaviours. The following are the common behaviours observed to divide the customer base.

  • Purchase frequency
  • Vintage
  • Days since last visit
  • Average bill value
  • No. of days between consecutive visits
  • No. of times contacted
  • No. of times responded
  • Loyalty features:
    -Points earned
    -Points redeemed
    -Points expired
  • Discount pattern
  • Purchase occasion
  • Online presence
  • Product-related features

Note: Each feature differs for different industries and there can be more features or observed behaviours depending on the industry. For example, if you create behavioural segmentation for a food chain company, then features like Dine in or Takeaway are also relevant.

Advantages of Behavioural Segmentation

Behavioural Segmentation allows an org to do the following.

  • Develop behavioural marketing campaigns.
  • Segment customers on the basis of purchasing habits and create different marketing campaigns to target habitual buyers (where the sales process is fast) and complex buyers (where the sales process is slow and will require more information and guidance for shoppers).
  • Look at the actions of a customer and use content mapping to create content that guides the customer. 
  • Identify and target the customers who have the highest level of engagement with the org.
  • Develop new marketing materials that focus on the benefits that are significant to current and prospective customers.
  • Map out the buyer’s journey.


Customer single view checkpoint

You have to create a customer single view before running the advanced segmentation notebook with all the features that the user might use as inputs for the segmentation.

Note: All the KPIs/features must be present at the customer level.

The following are the prerequisites of creating a customer single view.

  1. Keep only one unique identifier for each customer in the table.
  2. All the features must be numeric data types.
    Note: In the case of categorical data types, hot encode it and convert it to numerical data types.
  3. No. of visits or Frequency is a mandatory feature for the single view as the segmentation will only be done for customers who visited the org more than once.
    Note: A customer’s behaviour can only be determined based on more than one visit.

K-means Clustering

To configure K-means clustering, follow these steps.

  1. Specify the number of clusters K. Use the Elbow method or Silhouette Method to get the appropriate number of clusters. To learn more, click here.
  2. Initialize centroids by first shuffling the dataset and then randomly selecting K data points for the centroids without replacement.
  3. Keep repeating until there is no change to the centroids or the data points assigned to clusters.
    Note: To learn more, read Customer Segmentation Using K Means Clustering.

Input parameters

You need to provide the following details before running the Notebook.

Input ParamDescription
Single View DBEnter the name of the database which has the customer single view table.
Single View Table NameEnter the name of the table of customer single view.
Frequency ColumnEnter the name of the column  that has information related to no. of visits made by the customers.
User IdentifierEnter the name of the column which has the unique identifier for each customer.

Advance Customer Segmentation Data Flow

The following is the flow of Advance Customer Segmentation.

  1. Null Values Removal: If a certain feature has more than 95% of its values as null, then that feature is dropped from the analysis.
    If a certain feature has less than or 20% of its values as null, then the null value is replaced with the median value. In other cases, null value is replaced with 0. 
  2. Scaling the Feature Matrix: In order to bring all the variables to a common scale, you have to transform the feature matrix and bring it to common scale.
  3. Feature Reduction: The single view consists of many features and some features will correlate to other features. To avoid this, a famous dimensionality reduction technique named PCA is used. 
  4. Selecting the right no. of Principal Components(PCs): The principal components reduce the dimensions by explaining as many variations as possible. Right now, the cut off of explained variance is 80%, which means only the number of principal components till which the cumulative explained variance is <=80% is considered.
  5. Creating Segments: The K Means clustering method is used for creating the final set of segments after reducing dimensionality.
  6. Final Output: Get the final output of the notebook from command 35. After interpreting the output, the user can run the notebook again and input the no. of segments to be created in command 37.

Notebook Links

Open your cluster-specific link provided for the Notebook.

Notebook
Cluster links
Advanced Segmentation
India, SEA, EMEA