Retailers often group customers into segments, ignoring the fact that one customer in a particular segment might like certain items and another might hate them. This is why retailers are increasingly using machine learning to personalize not within segments, but within themes shaped by each customer’s unique tastes, aspirations and preferences.

 

James Glover, co-founder and CEO, Coherent Path

James Glover, co-founder and CEO, Coherent Path

Every email marketer recognizes that a cost-effective campaign must be targeted toward specific customers. In the past, many brands achieved a limited degree of targeting by segmenting their audiences—grouping their customers into “personas” based on their ages, genders, locations and other attributes.

While audience segmentation worked effectively enough throughout the early 2000s, today’s customers require a far sharper degree of personalization to prevent them from hitting the “unsubscribe” button—let alone make an actual purchase.

Every high heel-related creative you send to the uninterested group of women generates zero ROI and is a wasted opportunity to offer engaging content.

Marketers now recognize that lifelong customer relationships demand personalization at the individual level—and that demands a degree of message and creative tailoring that segmentation can’t deliver. In fact, the only way to achieve such fine-grained personalization at scale is to use a machine-learning system that intelligently crafts a unique email campaign for every subscriber.

advertisement

Here are three key reasons why brands are moving beyond segmentation, and using machine-learning personalization to personalize deeper within a theme.

Reason 1: Customers who like the same kinds of products rarely like the same exact products.

Most email campaigns begin with a strategic objective—typically to increase sales of products in a certain category. In the past, marketers often designed artificial customer personas around these sales goals, then segmented their audiences and targeted their campaigns accordingly. But this approach carries significant risk, because many customers with similar shopping habits may actively dislike similar products.

Say, for example, that you’re crafting a campaign to market your autumn shoe collection. You might be able to group 40 percent of your subscribers into the segment of “women who buy shoes for the office” and build an email using pre-selected creatives that include high heel shoes. Most likely some of those women hate wearing high heels to their jobs—while others buy only high heels for work, and ignore all other shoe-related creatives you send them.

Every high heel-related creative you send to the uninterested group of women generates zero ROI and is a wasted opportunity to offer engaging content. And that’s just one reason why retailers are now advancing beyond just segmenting at such an out-of-focus level of the product hierarchy (work shoes for women, in this example) and are instead using sophisticated models to personalize themes at the individual level.

Reason 2: Thematic personalization provides much more relevance than traditional segmentation.

For outdoor retailers, every season offers a change in wilderness activities—creating a whole range of opportunities to switch up their campaigns. In the past, most outdoor marketers would simply create seasonal customer segments around artificial audience personas—the retired fisherman, the hiking couple, and so on—and send the same series of emails to every customer in a given segment.

advertisement

As the shoe example above demonstrates, however, two customers in the same segment may actively dislike products assigned to that segment—which means every irrelevant creative sent to those customers is a creative wasted. This is why retailers are increasingly using machine learning to personalize not within segments, but within themes shaped by each customer’s unique tastes, aspirations and preferences.

For example, some customers in the outdated “retired fisherman” segment might prefer quiet fly fishing in mountain streams—while others might be deep sea anglers. Instead of trying to shoehorn both these groups into the same customer segment (to which they clearly don’t belong), a machine-learning system can start from the theme of “fishing,” then design a unique campaign for each customer based on their individual interests.

Reason 3: Even customers with similar taste will respond to the same creative at different times, in different contexts.

Although many customers have sharply contrasting taste in products, many will (of course) be interested in exactly the same items at some point. Even so, they won’t all want to buy those products at the same time, or in the same context.

How can you know when each customer is most likely to make their purchase? Without machine learning, it’s an impossible task—but with the help of an automated system that learns from each interaction with every customer, it’s easy to package each creative in the way that’s most likely to generate a sale, and send every email at the exact moment each customer is most likely to click.

In fact, machine-learning personalization does far more than just generate more ROI from each email—it turns every creative asset into a recyclable piece of content. By combining and recombining every creative you’ve approved within a given theme and template, the software can generate millions of never-before seen emails—and keep on generating them from one day to the next.

advertisement

That means every subscriber on your list will get exposed to the offers they’re most likely to respond to—all without any additional investment of time or resources on your part. And by allowing the software to personalize emails within a flexible theme, rather than a rigid and artificial customer segment, you’re vastly increasing the odds that you’ll connect with every customer on a personal level, fostering relationships that last for life.

Coherent Path provides predictive analytics software designed to surface products and categories that meet consumers’ evolving needs over time.

 

Favorite