Segmentation is more complicated than ever before. The number of meaningful, actionable segments has grown from hundreds to thousands. Machine learning is needed to handle this volume of segments. According to Eric Bosco in his article “A Smorgasbord of Segments” in MarketingDaily, marketers are in the dark about how many segments there really are, and how fine-grained their analysis will need to be in order to effectively reach segments that actually matter. They need a higher-resolution model of their audience in order to make relevant offers.
Bosco’s favorite “parlor trick” is to ask audiences at his speaking events to guess at the number of online advertising audience segments in the realm of digital marketing. They usually guess between 200 and 500, when the answer is closer to 70,000. But do we really need to address all those segments to get our marketing segmentation right? Which segments are important?
Bosco argues that we’ll never figure out the answer to those questions unless we do three things:
- Move away from manual: Automation and machine learning are key to find the audience segments that are meaningful to your business.
- Use current segments as a benchmark: Sometimes it’s better to delve deeper into the segments you already have and uncover previously unseen behaviors within that group. You’ll still need to pay close attention to their online habits.
- Continue evaluating segments: “Factors like weather, time of year/holidays, social trends and the economic factors all alter online behavior,” Bosco writes. Never write off the potential of external events to affect the size and shape of your audience.
Bosco concludes that the best way to approach each of these directives is to make use of new, powerful computing tools that have made data less of an obscure science of the few and turned it into revealing, actionable information for direct and immediate use by marketers. Only machine learning can truly make a meaningful dent in the vast field of demographics through which marketers now wade, helping them by filtering out irrelevant facts and presenting a narrower, more focused field of signals and actions for business users.