In this special guest feature, Patrick Smith, Manager of Data Science Services at Mather Economics, discusses the value of the ability to attach digital advertising revenue to the online activity at a customer level to show what content creates the highest revenue stream. It’s this knowledge that can (and has), in turn, help publishers create successful, more profitable business strategies. Patrick has participated in multiple projects varying from a valuation of federal wetland restoration to the creation of multiple econometric models estimating re-acquisition propensity. Additionally, he creates optimized renewal pricing strategies and performance reporting for Mather clients.
The longstanding practice of customer segmentation is experiencing a resurgence in the digital realm as a result of new insights into customer behavior combined with new and evolving ways to interact with the individual.
Why Segmentation?
At its base, segmentation is the practice of separating individuals into groups based on similarities. In the business sense, tailoring a marketing approach for particular segments based on behaviors or revealed preferences can shift customer interaction from a brief impersonal encounter to an experience that cultivates repeat business. This is no different in the digital arena.
Segmentation of online traffic is essentially a way to increase revenue yield by tailoring the visit to the individual based on prior behaviors. Using these characteristics, there are several methods a site can employ to achieve a revenue lift, one of which is targeted advertising. The ability to target specific visitors is highly valuable to advertisers and is something for which they’re willing to pay extra. Placing an ad in front of a visitor is important, but advertising to the right visitor is invaluable.
Another way to drive revenue is to increase engagement amongst visitors by elevating the user experience. The thought here is that a customized experience for the visitor based on prior behavior will encourage the individual to return to the site or sign up for premium access. For content producing sites, customizing content based on revealed preference encourages return visits which not only drives additional advertising revenue, but correlates to higher conversion rates for premium subscriptions/memberships if available. Additionally, understanding your visitor’s preferences can also be used to design targeted emails that may contain promotions, event advertising or specialized content to further engage your visitors.
That’s why creating segments is important, but how are they created?
How you segment is completely contingent on why you want to segment. Once goals are determined (increased CPMs, increased traffic, etc.), there are a few ways to create segments. The key is to keep it simple. Simplicity allows for clearly defined segments upon which predetermined business strategies can be executed.
In the vein of simplicity, much of segmentation in the digital world is merely distinguishing visitors who are engaged versus those who aren’t. It’s a fairly straightforward process to group visitors into engagement tiers based on recent activity, whether using an algorithmic approach (e.g. K-means or Hierarchical clustering) or using predetermined engagement thresholds (i.e. engaged visitors have more than five pageviews). Determining who is engaged is a sensible first step in the overall segmentation because it identifies visitors who are paying attention and those who may react favorably to custom treatment.
The chart here shows the results from a segmentation based purely on engagement metrics, where the end goal was to increase the number of engaged users and increase subscription conversion rates. Four tiers were created to determine a few levels of engagement, and the ultimate strategy was to customize the user’s experience in a way that would encourage movement up the engagement ladder.
While the goal for the middle tiers was to increase engagement, the goal for the top tier was to drive conversion. Segmentation in this case was integral not only in the identification of various levels engaged users, but also in illustrating the pipeline a visitor would need to navigate to go from less engaged to fully engaged.
This is just one example of the potential uses for segmentation in the digital world, and depending on the principal goals of the segmentation, there would be any number of ways to go about constructing the groups. It is, however, safe to say segmentation has found a new home full of possibilities in the digital world.
Sign up for the free insideAI News newsletter.