As a simplified definition of what they do, marketers strive to reach and influence the right people for their products or services, and look to constantly improve their strides in doing so. As a means to that end, most marketers are accustomed to thinking in terms of “segments.” It’s how we have always defined our audiences. By using demographic and lifestyle descriptors, we set the best possible definitions we can of our target audience and thereby create little compartments, sometimes hundreds of them, through which to market to our consumers. But, here is the problem: people are not segments, people are people, and people are dynamic and complicated.
Even if marketers have been slow to adopt the new options, as the science of defining and targeting audiences has advanced, so has the realization and evidence that segmentation alone is inadequate. This is especially true in mobile marketing, where the channel itself exposes an exponentially growing set of relevant consumer characteristics and behaviors to consider. Here’s an illustration of where and how things can snowball in the wrong direction when doing things without the benefit of today’s audience science – relying only on crude segmentation methodologies, something so many among us still knowingly do.
The Thought Process of Segmentation
First of all, we know that targeting everyone is not effective or efficient, so on the pretense of getting more “targeted,” we start by establishing age, then gender, and then income. As a layered snapshot, this feels like a step in the right direction, but it’s still a long way off.
So next, we make our segments tighter or more defined. We add in more about the person’s lifestyle. She’s a soccer mom or a golfer. At a glance, it’s better, but even with a target population like 26-35, high-income moms, with more than two kids, age 6 to 17, we have hundreds of thousands of U.S. consumers. This is still a very large, amorphous population, with plenty of attributes that might actually make them NOT a target for us.
Next, and here’s where the snowballing begins, we do our best to add interests, hobbies, or mindset. But, without today’s more advanced data models, how do we do that? Maybe we have third-party data based on something the person has done or purchased in the past, or because he or she filled out a survey and indicated interests. We also use context or media browsing, consumption behavior, looking at where our target audience is spending time. Context is a great indicator of not only interests, but tells a little bit about what the person is thinking about at that time. Context is bigger than any given website or digital path.
So, even when you add all of these things together, you still have large pockets of target consumers who may or may not be interested in your message, especially at a given point in time, based on various lifestyle factors or real-life contextual components. There’s got to be a better way.
Yet, what do we do? We keep adding more and more variables – characteristics, behaviors, contextual moments – to the mix, and we “refine” our segments even further. The result is hundreds of segments or marketing compartments, and maybe a better solution. Is “maybe” enough? No.
What Segments Create – Waste and Noise
For all intents and purposes, and for all your diligent layering of attributes, when you’re operating like this, you’re still using segments. And, there is a generally recognized understanding of segments that is: we “know” not everyone in the segment is going to prove a responsive or engaged recipient, and we are OK with that. When we use segments, we accept as a goal of the segment, and the refined more granular versions of it, that we will end up with a group of folks who is “more interested than not.”
Somehow that process of elimination feels like it will point to a higher response probability than if we use no segmentation at all. So through the process, we actually know segments deliver waste and noise, even if we find a way to reduce the waste a noise a bit as we move along.
It’s also worth noting that the more segments you create, the more you have to manage. A segment is only as good as the responsiveness it delivers, and your associated knowledge of that outcome, so you can build on or continue to seize the opportunity of that effectiveness. But the added labor of managing so many segments quickly overshadows the value of doing so. Not to mention that fact that people change segments, and you set yourself a bit of a trap by creating highly granular, tight segmentation, from which audiences will inevitably shift.
And, a final issue, what about the unaccounted for, though valuable, targets outside our segments? The outliers? The newly emerging? The exceptions to the rule? When we’re using pre-defined segments, how do we reach these folks? Or even know when they are there to be reached? This is an especially important consideration when we have a new product, offer, or channel with which we want to grow or expand our existing customer base.
The Solution = Segment of One
So how do we address these needs, while not ignoring the equally critical questions of measurement and scale? These are the things after all that we thought segments were solving for, in the first place.
I’d suggest an approach of focusing on a “segment of one” effectively resolves these issues. But how do you create and use segments of one? How and why does this process work?
Imagine if you could realistically manage 240 million different “segments,” roughly the number of U.S. consumers over the age of 18. What if you could continually evaluate your specific goals against each one of these “segments” and make a point in time decision to include them, or not, in a particular engagement moment or campaign? I’d argue you’d have a fairly effective targeting plan, and certainly better off than making similar decisions against eight, 15, 70, or more than 350 segments, but I’d also forecast you’d still be generating noise, waste, and missing best opportunities.
Why? Because when it comes to understanding who we are as consumers, it’s even more complex than who we are as people. The state of being a “consumer” encompasses engagement and purchase decisions that go well beyond who we are and/or what we like. My consumer decisions (actions), especially those related to my interests, willingness, or intent to engage in a marketing message or advertisement campaign, additionally involve my attention and access. Did I notice? Did I care to notice? Do I have the time? And can I do anything about it right now, or ever? Add to this the notion of a “mobile consumer,” and the complexity multiplies again, if for no other reason (though there are many) than because we as mobile device carrying consumers, are still learning and deciding how we are or eventually will use this device and channel to gather information, make decisions, and/or ultimately make purchases.
So, in effect, my pie-in-the-sky example of 240 million segments, is better imagined as 240 million times 10X every consumer times all the variables and continually shifting factors that impact their ultimate consumer decisions and actions!
Enter multi-variate logistic regression analysis. Yes, these are words that may ring a dim bell from that statistics class you couldn’t wait to be over, but it’s also the approach that allows us to accomplish the very idea expressed above. The ability to evaluate not only very granular consumer characteristics, interests, and behaviors, but to consider the combinations, interactions, and continually shifting nature of these factors that equally impact the end result – i.e., will targeting and engaging with this consumer, right now, right here, result in the outcome I’m trying to achieve with the marketing dollars I’m spending.
Add in Big Data (all the granular data noted above along with the ability to process it very quickly), along with machine learning (think rapid and automated analysis), and suddenly the ability to manage “240 million times 10X” becomes a reality. No longer do we need large aggregate imprecise segments to manage our ongoing targeting and analysis efforts. We can target one person, at a particular time and place, with a message tailored specifically to his or her needs and interest at that very moment, and we can do it again and again, in a systematic and measured way.
Header bidding is a programmatic technique that allows publishers to offer their inventory through multiple ad exchanges before they serve up ads from their ad server.
Whatever approach you take to your m-commerce project, one thing is certain: if you want it to deliver the results you’re expecting, context should be front and centre of your design.
As Facebook keeps changing its news feed algorithm, one constant factor is the domination of video content and so brands keep experimenting with ... read more
How are mobile payments, bitcoin, blockchain and other financial services technologies enhancing the consumer purchase journey?