What Good Is Data If You Don’t Use It?

Jun 19, 2017


A foundational criterion for becoming an Elicit employee is being a data omnivore. Whether you are a data scientist, technologist, marketing strategist, or account manager, you must eat, sleep, and breathe data. Some may call it an obsession, but we just call it fun.

The good news is that most people are starting to agree with us. We’re seeing companies collect a lot of data on their customers. The problem is, once they have that data, they aren’t sure of the best way to use it. According to an IBM research study, 80% of data is “dark and unstructured,” meaning it may not be available for the company to analyze or actually use because it’s in the wrong form or on a platform that doesn’t allow for it.

No matter what state your data is in, whether your organization is advanced in customer data capture, analytics, and application, or is just starting out on a customer insight journey, there are likely nuggets of customer data that you can use in your good data to build customer attributes.


A customer attribute is a descriptive or predictive measure of a customer’s behavior. You build customer attributes by aggregating and modeling the behaviors of customers from available data to summarize the most important, or most commonly utilized, components of customer behavior. Some basic examples include average order value, purchase frequency, and email open rate.

When you get ready to start building customer attributes, you’ll want to make sure you’re covering the nine buckets of customer behavior:

This list is extensive, but it’s important to collect and track a variety of customer behaviors to get a full picture of your customers, and how they are interacting with your business. For a customer attribute to be effective, you must be able to score almost every customer in your database on that metric and track changes over time. Once you’ve productionized your customer attributes you can start to really unlock their value by getting to know your customers better.


The beauty of customer attributes is that they have value at the individual customer level and in aggregate. Let’s take a basic attribute—purchase recency—as an example. You can track purchase recency at the customer level to inform targeted promotions and triggered email campaigns; at the segment level to understand how purchase recency varies across different groups of customers; and in aggregate as one component of the overall health of your customer portfolio.

While there are endless uses for your customer attributes, in this post I will focus on four immediate, high-value applications.


A good place to start using your customer attributes is to set goals, track progress, and report on results. A customer attribute should be treated as the source of truth for a business metric, and thereby immune to spin and formulation distortion. This is critical for creating consistency in how you define key metrics, set goals and KPIs, and track progress over time and across the organization.

If your customer attributes live in a customer data warehouse, or other OLAP environment, they can easily be piped in to business user tools like Microsoft Excel, Tableau, and RStudio. This enables your teams to quickly build and maintain reports and dashboards to track results over time.

A great place to start is with a customer portfolio health report. The report should measure the overall performance of the customers in your portfolio. You should include customer attributes from each of the nine buckets listed above in your report, and make sure to monitor changes in behavior over time. Once you build customer segments, you should incorporate them into your reporting as well to understand how different groups of customers perform and behave.


Customer attributes also serve as the basis for data analytics and predictive modeling. Building customer attributes makes it much easier for your data scientists to unearth customer insight gold, since they don’t have to dig through a bunch of dirt to get to what they need.

Customer attributes are mini-models of customer behavior in and of themselves. They can also be linked together in an infinite number of ways to build complex models, or to answer business questions as they arise.

By creating customer attributes, you also create efficiency and reduce the likelihood of human error in data science. If your teams can pull from a centralized metric repository, they don’t have to spend time navigating your data environment and starting from scratch every time they want to calculate an important metric or analyze customer behavior.


Campaign design, list selection, and testing should all be based on customer behavior. By now, I hope it’s obvious that if we’re talking about customer behavior, we’re talking about customer attributes. Once they’re productionized in your OLAP environment, the customer attributes can feed applications and software like email service providers, dynamic website functionality, and integrated marketing platforms, using operational data stores.

Let’s say you want to run a promotion on patio furniture. Using the customer attributes, it becomes easy to evaluate various options and simulate results. You decide to run a test on customers that are members of your loyalty program, have an average order value greater than $100, and have not purchased patio furniture from you in the past six months. Since these are all attributes in your customer data warehouse, you can easily pull test and control groups, calculate potential lift, deploy email and direct mail campaigns, and measure success.


Lastly, customer attributes are the building blocks for building customer segments. Customer segments are deeply nuanced and differentiated strategic customer groups. Customers in each segment exhibit similar behavior to the other customers in their segment, and markedly different behavior from customers in other segments.

Each segment is effectively a stack of customer attributes that are used to both define the segment and profile the customers in that segment. I could spend pages talking about the benefits of segmentation, but I will spare you and instead direct you to my colleague Chuck’s post on segmentation and Brooke’s post on the different types of segmentation if you’d like to learn more.


The bad news is, you will never be done building customer attributes. Your customers’ behaviors will change, your business priorities will shift, and your teams’ hunger for customer insight will evolve. That all translates to developing, updating, and archiving customer attributes. The good news is, once you start designing and leveraging customer attributes, you will want to keep building them.

I will leave you with one final tip: if you’re going to ask your customers to tell you about themselves, you better use that information. What’s more frustrating than having a company ask you to provide your name multiple times, or to never open a single email, but continue to receive them every day?

By building customer attributes it’s not only easier to learn about your customers, but also easier to show them that you’re listening.