Does your customer segmentation look more like channel segmentation?
- February 5, 2021
- By Maggie Wrona
At its core, customer segmentation is grouping your customers according to specific attributes in order to personalize their experience with your brand. This should result in more timely and relevant communication with customers at any point in their lifecycle and is the first step toward reaching every marketer’s goal: one-to-one personalization at scale. But as companies take their first steps towards this ultimate goal, often the segmentation tactics they deploy focus more on the individual activation channels, rather than customers.
Focus on customers, not channels
Customer segmentation has traditionally been a channel-specific initiative. The email marketing team may have one set of segments and audiences, the website marketing team another, etc. These segments may even have the same intent—e.g., identifying highly engaged customers—but use different, channel-specific data to define them. The email marketing team may define “highly engaged” as someone who has opened four emails in the last six weeks, while the website marketing team defines it as someone who has browsed more than five pages in the last month. This creates a fractured and inconsistent view of the customer across the business and prevents organizations from launching relevant omni-channel campaigns.
In fact, marketing technology thought leaders are recognizing that organizations must move from this channel-centric perspective to one that is focused on the customer experience.
Small gains from segmentation aren’t good enough
Marketing teams may recognize that their current customer segmentation models aren’t perfect. But they’re good enough to give them measurable lift, whether it’s improving open rates or driving more users to their online properties. While following these models might be familiar and comfortable, the end result can come with an unfortunate cost.
Most importantly, it can cause marketers to make false assumptions about customers. Someone whose engagement is shifting from email to online, for example, might look like a churning customer to the email team and an engaged customer to the web team. This can lead to mixed messages and higher marketing spend as channels “overmarket” to engaged customers.
A customer data platform can help segmentation
A customer data platform (CDP) can help begin to address these pitfalls. By bringing siloed data from different channels into a single view for customer-, rather than channel-based segmentation, marketing teams can gain several cross-channel efficiencies.
Let’s return to the first example of “highly engaged users.” With a CDP, both the web and email teams can now build out a shared definition of what a “highly engaged” user is, taking into account data points from both channels (and potentially even more!). They only have to build this audience once, centrally located inside the CDP. Then this one agreed-upon audience can be shared to web, email, social, and ad activation tools.
This results in a consistent user experience wherever a person interacts with your brand. Organizationally, it allows the marketing organization to start to plan and communicate, focusing on shared company goals. They can target their channel campaigns appropriately, making the most of their budgets.
Data science-based behavioral data yields more accurate segments
A CDP can also bring data science and machine learning to support customer segmentation, providing marketers with insights that may not be so obvious at first glance. Having a shared definition of what it means to be highly engaged is great, but it’s even better to have algorithms tracking your entire database of customers to identify and inform marketing teams which customers are engaged, which are likely to churn, which are disengaged, and which have a high affinity for low-heeled women’s boots–all without you having to lift a finger.
When using data science-powered behavioral data, you don’t have to define what it means to be engaged or not, or try to predict behavior by making educated guesses. The AI algorithms do it for you. For example, this month you may say five pageviews and three email opens define an engaged user. Yet with a holiday coming up, or some other new trend taking hold, suddenly a majority of customers may start visiting 15 or more pages. Using a CDP like Lytics, with its AI decision powered by behavioral scores, you won’t have to go back and adjust that definition for engaged users manually. It will be aware of the changing trends and update the definition of “engaged” automatically–and it will look beyond the obvious data points to create a richer, more robust definition.
These types of data science-based behavioral audiences open up new use cases for marketing teams. They can suppress engaged users from a new customer acquisition campaign in Facebook in real time with evergreen data. They could reduce the number of emails that an engaged online user receives to avoid over-marketing them. Marketing teams can pay special attention to users likely to churn with a special renewal offer four months away before their subscription expires.
Behavioral data helps remove bias
Returning to a previous point about making assumptions: As marketers, we often have to rely on third-party demographic data. Doing so exclusively can make us miss key opportunities and allow our own biases to get in the way of what each customer wants.
Let’s say that a fashion retailer is launching a new line of fashionable women’s shoes. Through demographic data or just plain event-based behavioral data, you may choose to target “women aged 25-54 with an income in excess of $75,000,” or “people who have viewed or purchased women’s shoes this year.” Aside from the fact that third-party data may not be totally reliable, this type of segmentation misses out on men who may wish to purchase women’s shoes for themselves or loved ones because of the assumption that only women of a certain age will want to make this purchase.
If we turn to data science-powered behavioral segmentation as a complement or alternative to the above, we can instead target anyone with an affinity for women’s shoes based on their behavior, or run predictive models using behavioral scores and more reliable first-party data to determine who out of your customer base is most likely to purchase a women’s shoe.
Tuning into the customer rather than the channel
To be truly effective, marketers need to escape the channel mindset. That doesn’t mean ignoring channel outcomes, since marketers will ultimately be measured on their own successes. Instead, it means recognizing that a cross-channel view of customers benefits every channel, whether the benefit is improving engagement or increasing return on advertising spend.
Marketers also need to become better at harvesting and leveraging first-party behavioral data in their customer segmentation. Behavioral scoring, predictive scoring, and finely tuned lookalike models all have a place in the modern marketer’s toolkit, particularly if they plan to move beyond the small, incremental improvements of the past. We are, all of us, ultimately segments of one, with behaviors that change even when our age group or income group doesn’t. Adapting to those changes and adjusting our customer interactions to match them consistently across channels is the only segment that really matters.
To learn more about why companies need to focus on customers above all else, check out this webinar featuring Lytics President Jascha Kaykas-Wolff and Wharton Professor Peter Fader.