Editorial
The Gist
- Enhance understanding. Customer data segmentation enhances insight into diverse customer needs and behaviors.
- Optimize strategies. Utilizing both attitudinal and behavioral data sharpens market approaches.
- Drive growth. Effective segmentation directly supports business innovation and customer satisfaction.
Customer-centricity is the process of continuously optimizing the fit between customer needs and providers’ products and services. Customer research in the form of interviews, surveys, analytics, data segmentation and more has the goal of discovering customer characteristics as well as stated and unstated needs.
As it turns out, in most cases the customer base is not comprised of a single homogenous group, but multiple segments that differ in aspects such as geography, income levels or feature needs. This provides an obvious challenge for the producer of the offering: Who should I market and sell to? Who should I design and develop for? How can I provide a product that is useful, usable, and desirable for my audience when different customer segments have different needs and requirements?
Customer data segmentation allows you to analyze if there are several distinct groups in your addressable market and what they are. It also allows you to prioritize the segments that you should address. After all, the goal is not to serve every possible segment, but rather those segments that will be best served by your offerings — those that allow for sustainable growth and profitability.
Both the identification and prioritization of user segments should be based on data rather than gut feelings. This is appropriate as a correct product-market fit has huge implications not only on the product itself, but also on other areas, for example, marketing: What message should we get out to whom via which channels at what points in time?
Below I have outlined different customer data segmentation approaches that teams can use to get the best outcome.
Data Segmentation Based on Attitudinal Data: Voice of the Customer
Directly engaging with customers and potential customers about their needs has always been a cornerstone of user-centered design. By assessing the voice of the customer (VOC) through methods like interviews, focus groups, or surveys, not only are requirements, wishes and needs identified, but it also becomes apparent that different customer segments have their own needs. For example, the users of a company’s customer portal may fall into two groups: corporate customers and end customers. They hopefully have the same overall intent to buy, but they also have distinct needs and expectations toward the portal.
What is nice about customer data segmentation through VOC is that it can be done early on in the product lifecycle before there may even be a product. This supports a true outside-in product strategy where you start with customer needs and then produce an offering that serves those needs.
Customer Needs Questions
Here are example VOC questions that explore general customer needs:
- What are the most critical challenges you face in your daily work?
- In an ideal world, what would the perfect workflow look like for what you do?
- What are some of the capabilities your dream product would provide to you?
VOC Questions
Specific to an existing product, VOC questions might be:
- Are there specific features or functionalities you find lacking in this product?
- Have you experienced any pain points or frustrations while using this product?
- Which features of this product do you find most valuable for your work — and why?
With regard to data segmentation, it is important to also assess demographic information from those being interviewed or surveyed. This may include things like age group, education level, household income, gender, place of residence, job title and industry. Combining the answers from the VOC content questions with the demographic information gives you the segments:
- Who (demographics) said what (answers to VOC content questions)?
- Who (demographics) needs what (customer abilities)?
There are statistical (e.g., cluster analysis, to be covered in more detail below) and non-statistical (e.g., affinity diagramming) techniques to facilitate this. The prioritization of customer data segments can be based on:
- The segment size in terms of customer volume or revenue potential.
- Your capability and capacity to serve the required needs.
- The size of market share you can seize.
How many customers do you need to engage for your data segmentation? Ten interviews will hardly be representative of your addressable market. With surveys you have a much wider reach although the depth of insights will be lower. A staggered combination of interviews and surveys can do the trick. Interviews can effectively identify first-order segments that then can be further explored and validated through high-numbered surveys. Conversely, interviews can be used to understand nuances and details about segments that were identified through surveys.
Related Article: 5 Ways Data-Driven Insights Are Reshaping Customer Segmentation
Customer Data Segmentation Based on Behavioral Data
Behavioral data is often preferred over attitudinal data for customer research because it is based on what customers actually do (as opposed to what they say they did or would do), which is less biased and has greater predictive power.
Observational Voice of the Customer (OVOC) is a method that goes beyond VOC in that it involves gathering insights from customers through observing their behaviors and interactions in their authentic work environment. This provides richer data and deeper insights, thus can provide better-grounded segments.
Learning Opportunities
Conference
May
22
Salesforce's Connections Chicago 2024
Register
Conference
Jun
3
Customer Contact Week Las Vegas 2024
Register
Analytics-based segmentation approaches assume that there already is a product and users who engage with it. These users produce data that we can mine and derive insights from without the need to ask them questions. With that, these methods lend themselves well for establishing and validating customer data segments, and to course-correcting a product’s feature set when necessary.
Demographics Analysis
In addition to the information that can be gathered in VOC, analytics platforms like Google Analytics and techniques like UTM parameter tracking and IP address tracking can provide additional data helpful for defining customer segments:
- Technology: Platform (phone app, web app, PC client), form factor (phone, tablet, notebook, desktop computer) operating system (iOS, Android, Windows, macOS), browser (Chrome, IE, Firefox, etc.) and browser version
- Marketing Channel: For a web-based product — how did users arrive at your site, e.g., directly, organically from a search engine, through a referral, from online ads? If users came from online ads, were those search ads or display ads?
User Behavior Analysis
Through analytics platforms (e.g., Google Analytics) and session replay software (e.g., Microsoft Clarity) you can track what users do in your product on a granular level down to a single mouse click. Some of the most interesting data segmentation metrics are:
- Number of new users: The count of users who started to use your product.
- Retention: The percentage of users who still use your product X number of days after starting.
- Feature Engagement: The percentage of users who utilize certain features of your product.
- Core Events Execution: The percentage of users who carry out user actions that you consider as central or important in your product, e.g., adding an item to the shopping cart on an ecommerce site.
Combining demographic data and user behavior data allows you to derive customer data segments together with their importance, e.g.:
- We gain 70% of new users from online ads, and 30% organically.
- Web users have a seven-day retention rate twice as high as mobile app users.
- Users from the aerospace industry utilize product feature A 3X more often than feature B, while users from the automotive industry use feature B 5X more often than feature A.
- Users who add clothing products into the shopping cart end up finalizing the purchase 66% of the time while users who add computer products only purchase 59% of the time.
Regression Analysis
This technique allows you to model and quantify the relationship between input variables and a target variable. For example, you might want to know the impact of certain demographic user attributes like industry or country on the retention rate. The one complicating factor is that categorical data like industry and country information must be transformed into numerical data.
The resulting model can be used for predictions, but in the context of customer data segmentation the real value is that it can tell you how much percent each of the input variables contribute to the target variable. For example, a result may say that a user’s industry contributes 70% to the retention rate while country only contributes 30%.
Cluster Analysis
This technique categorizes data points into groups (clusters) based on the similarity of the individual data points. What constitutes similarity depends on your data:
- For numerical data (like the retention rate of users), the Euclidean distance is used for measuring the similarity (“KMeans clustering”): How far are two data points apart in the n-dimensional space that is formed by the number of input variables?
- For categorical data (like the country of users), the frequency of occurrence of each data point can be used as a similarity measure (“KModes clustering”).
A significant disadvantage of clustering is that the resulting clusters are unlabeled which makes it challenging to intuitively understand the meaning or significance of each cluster. Additional effort is then required to analyze and interpret the characteristics of each cluster.
Social Network Analysis
This is interesting for the increasingly popular products that are social and collaborative in nature. With network graph tools like NodeXL Pro or Gephi you can represent and analyze the relationships and interactions between users. For example, you may want to understand the level of collaboration that users in an organization demonstrate in your product. Who communicates with whom? Who shares assets with whom? How often do users carry out collaborative actions? Network analysis gives you statistics that you can then use to compare the collaboration levels of companies, or departments within companies, belonging to certain segments.
In a network graph, there are elements visualized as nodes, and connections visualized as lines between nodes. Let’s say the elements are individual users, and a connection is a relationship between two users that is established through a collaborative action like sharing an asset with another user within a software tool. Here are some of the most helpful statistics that a network graph can provide:
- Network Density: How many connections exist among the users relative to the total number of possible connections? The closer to 1 the better as it represents a perfect network where all users are connected to each other.
- Network Diameter: What’s the maximum number of direct or indirect connections between any pair of users? The lower the better, indicating closely connected users.
- Degree Centrality: What’s the number of connections that each user has? The higher the better. Those users with the highest degree of centrality are considered “hubs” in your network.
- Average Degree Centrality: What’s the average number of connections per user, i.e., how many other users does the average user collaborate with? The higher the better.
- Eigenvector Centrality: Who are the users with a high degree centrality who connect to other users with a high degree centrality? This allows you to identify “influencers” in the network.
- Betweenness Centrality: Who are the users who connect sub-networks? For example, users in the marketing department collaborate with each other, and so do users within the HR department. The CEO of the company collaborates with the two department heads, thus connecting the two sub-networks. Users with a high betweenness centrality are considered “brokers” in your network.
Related Article: What Defines Customer Data Today?
Final Thoughts
Customer data segmentation approaches offer a powerful means for businesses to enhance their understanding of customer needs, behaviors and preferences. As always, a combination of approaches yields the best results that you then can utilize to create targeted marketing strategies, tailored product offerings and optimal customer experiences. This in turn positions you well for growth and innovation.
Learn how you can join our contributor community.
