Customer segmentation — A powerful tool to increase your marketing ROI

What is customer segmentation?

The formal definition: Customer segmentation is the practice of dividing a customer base into groups of individuals that are similar in specific ways relevant to marketing, such as age, gender, interests and spending habits.

They can usually be classified as:

• Demographic — Age, or gender
• Geographic — Location, region, climate, or market size
• Behavioral — Usage, recency, or propensity

Why is customer segmentation so important?

Segmenting your customers allows companies to effectively spend marketing dollars to yield more optimal results with higher returns. Companies can create specific marketing treatments for different segments and optimize them based on the data. This in turn will lower your marketing cost, increase retention and enable you to identify your best customers.

How does machine learning help you segment your customer?

One of the most powerful yet simple tools to segment your customers is using K-Means, which is an unsupervised machine learning clustering algorithm. What K-Means does, is it creates segments based on the data you provided, i.e. clustering without your supervision. It will scan through all information associated with each customer and learn the best way to group them together. The algorithm creates random centroids in your data set based on the number of clusters provided and slowly try to fit each data point into the closest centroid. The centroids will adjust automatically based on the last nearest datapoint allocated to them. This iterative process continues till all the data points are associated with either of the centroids hence creating distinct clusters. A simple image below can explain this process effectively.

I will now demonstrate how to successfully segment your customers using a great tool called KNIME Analytics using a mock data set.

This data contains the annual income of customers and their annual spend. Let’s use KNIME to first visualize the data and then use K-Means clustering to create 5 clusters in the dataset.

KNIME Workflow to create clusters using K-Means

Regular Scatter Plot of the dataset X Axis :Income & Y Axis : Spend

Scatter Plot of the dataset with the assigned clusters X Axis :Income & Y Axis : Spend

The image above shows the distinct clusters which would allow us to make more effective marketing strategies when we target these customers.

Analyzing these clusters gives us more insights about their behavior and patterns.

Findings

• Cluster 2 over indexes in average spend inspite of being in the lowest income group
• Cluster 5 and cluster 0 are the lowest spenders
• Cluster 1 represents high spend and extremely high income

Based on these above finding you can tailor specific marketing campaigns to these groups to drive incremental revenue or increase retention.

Conclusion

As demonstrated K-Means is a powerful tool for segmentation and can be extremely helpful for developing intelligent marketing strategies for your organization.