ContactMost of us grew up believing we’re special. Our moms, our teachers, and Mr. Rogers raised us to think that we’re one-of-a-kind. Now marketers have joined this choir of positive reinforcement, celebrating customer individuality by ushering in a new era of content personalization. Personalization, they say during their keynote speeches, will be the prime driver of marketing success within five years.
But data scientists are continuing to present case studies showing that customers are essentially predictable. Invest in technology and skills that collect and analyze customer data, they say, and you’ll find that people reliably behave like one another. Unique individuals? Those are called outliers. The rest of us unknowingly travel in packs, or segments, exhibiting wonderfully ordinary behavior no matter how special we think we are.
So, what’s the deal here? Are we unique, or do we predictably behave like others?
The promise of personalization
Outside of antiquated one-size-fits-all advertising, marketers largely rely on rule-based methods to classify customers into meaningful groups. Modern segmentation fundamentals remain mostly unchanged from Wendell R. Smith’s first formal description of it back in 1956: we still analyze customer data to extract patterns and relationships that underpin segmentation strategies, ranging from basic demographic groupings to complex model assignments. The success and longevity of this approach provides us with some evidence that we behave like others.
But what if the uniformity we’ve always observed in groups is nothing more than a collection of blurry customer snapshots? If recent technological advances dial up our resolution—much like Van Leeuwenhoek’s invention of the microscope—we may be on the verge of revealing a universe of detailed activity on a hitherto unremarkable surface.
With its promising of hefty gains, is micro-segmentation the next frontier of marketing? More importantly, is it right for your business?
Welcome back to Westworld
Personalization is trending. But, like many trends, it isn’t as straightforward as it first seems. Consider the diagram below, illustrating the spectrum of personalization:
Like any data-driven technique, personalization is a complex journey requiring not only the right technical acumen, but the wherewithal to recognize a critical point of diminishing return.
Applying a parameter value like “Welcome {CustomerName}!” to a web experience is technically personalization, but it’s barely a step up from slapping an invoice on a box before shipping it. When we add more data points, parameters, conditions, rules, and responsiveness to our manufacturing process, we progress along our personalization scale towards an upper boundary of theoretical supercomputers that out-finesse the softest human touch.
Traveling along the scale is costly. Although AI is rapidly progressing towards addressing our emotional nuances and idiosyncrasies, super-hot Westworld automatons are still a few years out—and humans still enjoy a healthy advantage when it comes to delivering bespoke products and services. Alternatively, typical cost-effective digital personalization solutions that fail the Turing test are restricted to the left end of our spectrum, where value is realized through modest lifts in performance at scale.
Don’t pursue personalization at all costs
This common form of digital “personalization at scale” is in many ways an oxymoron, or a misnomer at best. At its core, it remains a form of mass production—an efficient way to manufacture standardized goods. Our digital factories process raw materials of content and data assets using machinery built out of algorithms to rapidly manufacture online experiences. Though can we infinitely recombine these assets into “unique” but similar experiences, we ultimately strive for certainty towards a finite set of desired standardized outcomes.
Consider the following thought experiment: after millions of “personalized” iterations of a media campaign selling a collection of winter parkas, we find that people best respond to a small set of generic offers featuring visually appealing content. Should we invest in crafting more personalized content at the risk of current performance? No! There is no imperative for us to build and serve content with the finest granularity of personalization, as these algorithms function best by promoting content that works to the right people.
Blindly pursuing a vision of personalization can steer us away from scientifically arriving at other cost-effective ways of achieving our finite set of desired outcomes. All businesses can cross the point where personalization yields diminishing returns. So, before we chase after personalization because we’re told to, we need to get our fundamentals right.
Rules are the new segments
Tech allows us to almost instantaneously build responsive custom digital experiences from blueprints founded on complex, rules-derived aggregate data analysis. Thought your Netflix recommendations come from your viewing history alone? Hardly! Amazon recos? They literally say, “people also bought”! Variable insurance pricing? It’s from the analysis of many accidents, requiring statistically significant samples to produce multipliers.
We can be fooled into believing there is real intelligence underneath these recommendations, because frankly it feels pretty comforting when Alexa speaks gently about today’s weather. This magic trick of AI is what tech visionary Jaron Lanier calls “mostly theatrics with a little bit of science”.
We achieve this magic by experimenting with and measuring various tactics to learn what combinations of content work best for what types of people. By analyzing our data in aggregate to find reliable relationships that define rules that instruct computers, we come back to the tried-and-true fundamentals of statistical modeling, because rules are the new segments.
So, what should you do?
It’s tempting to detour the basics and dive straight into fancy CDP, DMP, and Journey Orchestration tools—and there are a lot of vendors out there who are beckoning you into the pool. These incredibly powerful (and expensive) platforms require premium fuel to run as intended, which is a set of rules and conditions applied to users. Might your collection of rules constitute a personalization strategy? Potentially, but not necessarily. You must consider your industry, your product types, your customer expectations, your budget constraints, your activation capabilities, and more. It all starts with understanding and trusting your data, spending time with it, building creative hypotheses, testing and validating them, and developing a plan.
In other words, it’s just like those marketing gurus keep telling us: one size doesn’t fit all, and content personalization isn’t for everyone. But is it for you? We can answer that question. With a realistic cost-benefit assessment of your personalization goals—and detailed insights into the risks of not pursuing it—we’ll reveal exactly how personal you should get. Let’s get started.