11436 SSO

How to Create Personalized Experiences with Behavior-based Segmentation

yongkim
Apr 16, 2015

Customer segmentation is a fundamental method marketers use to understand their customers and deliver personalized experiences. It’s used to group a large customer population into subgroups with similar attributes, preferences, and profiles. I’ve taken many approaches to segmentation, but what’s been really fun is demonstrating to our potential customers the behavior-based segments based on customer interaction data in Apigee Insights, our big data predictive analytics platform.

We’ve been able to demonstrate how adding behavior-based segments enhances the ability to personalize interactions with customers. Insights users can target their customers even more precisely based on recent behavior that indicates interest, or on the sequences of interactions that caused dropoff.

Segmentation strategies

There are many approaches to segmentation. Some are very basic, like RFM (recency, frequency, monetary), a segmentation strategy based on the recency of purchase or visit activity, the frequency of activity, and the money spent during a set period.

Business intelligence tools provide a granular method of segmentation, based on multiple dimensions such as age, income, and gender. Unsupervised machine learning algorithms such as K-means or hierarchical clustering algorithms offer an even more sophisticated approach, but it’s combination of art and science because there’s really no “right” answer in terms of the correct number of segments or clusters, or even the right set of dimensions or features to cluster by.

These customer segments/clusters provide the marketer a basis for personalizing offers or messaging targeted for each segment.

How to improve personalization

On one hand, these segments, whether generated by BI applications or machine learning methods, are very helpful in characterizing and describing the various types of customers in a database. However, they aren’t particularly helpful in guiding a marketing team on when particular groups of customers should be contacted.

For example, during a recent meeting with an online travel site, one of the challenges we heard about was the difficulty in identifying visitors that show “serious buying signals” within the customer segments the company had created. Identifying this specific group would enable “remarketing” with a personalized marketing message.

Insights Segment Manager

Even with a limited dataset (two weeks of session data) with high-level event data (consisting of logins, purchases, purchase failures, and cancellations), using Insights Segment Manager we were able to identify a recent group of customers that logged on, then booked a room, then tried to book another room, but failed, then dropped off. Combining this recent behavior pattern with the existing customer segments enables the travel company to have more personal, relevant, and impactful interactions with their customers.

With Apigee Insights, marketers can start with their customer segments, then explore journeys and identify behavior segments that have indicated interest, but haven’t purchased. Below is an example of one such customer journey:

Another approach is to start with a specific journey, then segment further based on profile attributes, or apply existing segmentation on top of the behavior-based segment. Here’s an example: we start with a specific journey consisting of a customer service call for information, then a store visit, then a web visit showing interest in the baby product category:

 

Once the journey is identified, Insights Segment Manager provides the ability to further apply additional profile attributes, such as age and gender, as well as predictive attributes, such as the likelihood to respond to a mobile-based offer.

Finally, integration with Apigee Edge provides a consumption layer for rapid exposure and integration with existing CRM applications via APIs. Below is a screenshot showing how to export the behavior-based segments to API BaaS to make them accessible via APIs:

 

Whether  your company’s segmentation strategy is BI-based or machine learning-based, applying an additional layer of behavior-based segments is a powerful way to complement and enhance your current understanding of your customers and improve personalization. You can try this out yourself by visiting https://accounts.apigee.com and signing up for an Apigee account. To discuss behavior-based segmentation and more, visit the Apigee Community’s Insights forum.

Microservices Done Right

Next Steps

 
 

Resources Gallery

News