In today's post, I am going to discuss an example of customer segmentation. In particular, we will look at an example from my political work and how two key concepts: identity and action can inform your customer modeling and segmentation strategy.
A Background Story
Years ago I was working as the "data guru" for a political campaign. After some investigation into my client's database of voters, his campaign manager and I decided on a simple strategy: instead of trying to persuade voters, target his most likely voters with a series of direct mail pieces designed to get them to the polls. This strategy was certainly not cutting edge, but my methods for identifying my client's most likely voters would be.
But then the campaign manager and I were thrown for a surprise: our candidate was insisting that, instead of our proposed series of mail pieces sent to his most likely voters, he wanted to spend nearly his entire campaign budget on just one piece to every voter in his district. His reasoning ws sincere: he wanted to be the representative for all of his district, and not just his potential voters. In other words, he wanted to educate his entire district about his vision. Consequently my targeting services would not be necessary.
Luckily his campaign manager was able to point out the folly in his reasoning: that if he wasn't elected, his vision for his district would never become a reality, and his communication to the entire district would be a waste. The candidate changed his mind. Eventually I would run several predictive models (a few regressions using past voter turnout and election results) and created a targeted list for his multi-wave direct mail campaign. He won his re-election by several percentage points. I like to think I was responsible for a few of those percentage points.
This episode illustrates a critical rule for marketing, which I use consistently, from political campaigns to fund-raising to traditional marketing: be very aware of who you are targeting and the costs of doing so. Seems obvious, right? But I am continually surprised by how this simple principle can get lost in a mire of data and strategy. Perhaps a vocabulary of identity and action can help.
Identity and Action: Understanding the Difference
Let's break up marketing efforts into two categories: identity and action. As a marketer, and this applies to traditional marketers, fund-raisers as well as political campaigns, you generally have two missions: either drive a message, which I call an identity, or drive an action.
With identity, we are attempting to either educate the public about our brand or product. But identities, your brand identity or your consumer identity, are not very malleable. They are hard, fixed preconceptions about the world around us. Think of my client the politician: trying to convince someone who does not already agree with his political position to switch their view-point would be extremely difficult, and in some cases practically impossible. How many mail-pieces and phone calls would he need to make in order to change a voter's mind? Would they ever change their minds?
The same problem exists for product marketers and fund-raisers as well: If someone already has a negative view of your product, how many marketing dollars would it cost to convince them otherwise? That is not to say that the establishment of your brand's identity should not be a part of your marketing strategy. Rather, have a clear head when it comes to establishing your brand's identity, and protect it fiercely. And know that altering it is costly. The same is true for products new to the marketplace: it takes an extraordinarily large effort to build brand a identity.
Action is different from identity. Action simply refers to the portion of the population willing to do something, such as make a new product purchase. In our political analogy, this is the voter that is going to vote in the upcoming election: regardless of their view of a candidate, you know that they are going to show up to the polls. Different from identity, our impulse to action can be sudden and capricious, whereas our identity is much more hard-wired. Think of the act of buying a new computer: I decide that my current laptop doesn't have the battery life I need, and just like that I am in the market for a new one. But which one I purchase might be a deeper question, summoning a deeper question of identity: am I a Mac or a PC?
The Data Scientist: Time to Break it Down
What does all of this mean for the Data Scientist? In certain obvious situations, it may mean you build two different predictive models, for two different types of behaviors. Once done, you can then analyze the interplay of the two. Think in terms of a Venn diagram: what is the intersection of the population that both have a positive view of my brand, and are in the market for my product.
Your Venn diagram could be shifted one way or the other: perhaps there is a strong need for your product, but you lack brand identity. Seek to quantify these numbers in your analysis. The intersection of the two is your obvious sweet spot, but also think in terms of how you might grow one or both circles.
Another way of interpreting this distinction is by looking at it in terms of degree (perhaps binned model scores). You may choose to spend less of your marketing budget on those consumers that score highest on both scales of identity and action. That is, you could think of them as already being at the tipping point, give them a gentle nudge, and instead concentrate your efforts on those that need stronger push.
Take your "positive identity" universe and segment those within in it based on their likeliness to take action and develop your marketing spend based on these categories, perhaps giving the greatest proportion to the "somewhat likely" category. Of course, your categories will most likely be more sophisticated than what I have laid out here, but this may give you a paradigm for communicating your analysis to your key stake-holders.
Take a look at the four quadrants in the graphic above. Which quadrant how might you treat each quadrant? Which ones should you put little or no effort into and which ones can be won?
Hopefully I have given you one another way you might think of your marketing universe and a tool for dissecting your analysis.
What are your thoughts? Was there a time when knowing the difference between identity and action was important to one of your campaigns? If so, I would like to hear about it.