When the focus is on metrics or statistics, there are really three things that marketers must know to use them effectively.
#1: Observed and Expected
Statistics typically are based on a simple equation: “Observed minus Expected.”
Ex 1: Comparing the target audience (the Observed) against the total population (the control or Expected). High or low indices can inform marketing strategies and tactics.
Ex 2: Assessing company sales performance, where the actual sales (Observed) are compared against prior sales (Expected).
Even as statistics get more complex, we see this. An “average” (aka “mean”) is another way of saying “Expected.” Conjoint (trade-off) analysis, a complex multivariate technique, works under the assumption that at the start, all attributes have an equal chance of being selected.
Most marketers don’t know it, but they “talk” in “Observed minus Expected” all the time: “If we do A, I hope to see B change by X.” Measuring that goal becomes the foundation for designing research, for the statistics used in analysis, and for seeking unexpected insights.
(For those of us into social media/discourse analysis and data mining, a special note: there often isn’t an “Expected” with these methods, which is one reason why they’re harder to analyze. The goal, moving forward, is to start establishing benchmarks against which findings can be assessed – or it will always be difficult to separate new, critical insights from the undefined norm.)
#2: Correlation vs. Causation
Because two things are related doesn’t mean that one thing caused another. We know that over-eating (unfortunately) causes most people to gain weight. However, most smokers – some say 90% of them – don’t get lung cancer. Smoking doesn’t cause cancer, but it’s clearly a (correlated) high risk behavior because many more of those who smoke will get ill when compared with those who don’t.
How does this impact marketers? You can’t really assume that doing A will cause B to happen. However, business models built on connections between operations, consumer attitudes, behaviors, etc. are very powerful in finding ways to reach business goals.
#3: Compound Findings
Independent facts can’t be compounded or strung together to simplify the “story” being shared with others.
Does this example tell us that our target should be women, age 25-34, who participate in outdoor sports? Not necessarily.
Assuming a perfectly distributed sample, we’d estimate that the target is 25% of our population (65% x 65% x 60%). But what if all the men – this is a hypothetical! – said they were into outdoor sports? That would mean that only 25% of the women said the same – reducing our estimated target size to 11% (65% x 65% x 25%).
While compounding or grouping results makes for a clearer story, marketers need to know that the story may not be accurate.
I recently came across the following quote from Carla Hendra, CEO of Ogilvy & Mather North America: “CMOs are under increasing pressure to deliver business results and to demonstrate the contribution that marketing makes to their organization. The days of guess work and soft metrics are over — Math Marketing is the future.”
A couple of reactions to the quote:
First, not everything is measurable. Love isn’t, apathy isn’t – just to name two.
Second, trends can’t always be predicted, and needs can’t always be anticipated, based upon what currently exists and is measurable.
I have no issue with being accountable. But I do have an issue with fear (in the guise of accountability) killing creativity. That would be a little like agency creatives only developing ads to beat the norms in copy testing; this has only ever produced average or slightly above average ads. I’m convinced that the original Apple “1984″ ad would not have tested well to norms.
I would really dislike seeing CMOs turn into “CFOs in new clothes.” I do understand the digital environment affords marketers more avenues for reaching micro-targets, and I think it and its arsenal of new tools is a wonderful thing. My primary concern is that the more CMOs are into the numbers, the greater the risk they won’t see the complexity of the people they’re trying to reach.
Early in my career, while working at Foote, Cone & Belding, I was called to jury duty in downtown Los Angeles. The case took several days and I got to know some of my fellow jurors. One – I can still see him to this day – was an African American man, about 60 years of age, who very tall and thin, and had spent his working career as a highway-sign laminator (and eventually supervisor) for the state. He had six grown children, all of whom he’d put through college; he himself hadn’t finished high school. To this day, he remains one of the wisest men I’ve ever met. Demographically, he was easy to categorize: older, less-well educated, African American male. Yet, even at 24, I knew that I had learned a life-long lesson from him: if I’d only seen him demographically, from my ivory tower, I would have missed the richness of his life and what I could learn from him.
Ultimately, what we learn when in relationships (yes, even with customers), makes for stronger relationships. My hope is that Math Marketing doesn’t leave the consumer too far removed from the equation.
Your thoughts?
In the past week, a new answer/search engine, Wolfram|Alpha, was launched. Brought to us from the folks that make Mathematica software, the website declares it is ”Making the world’s knowledge computable [in our] ambitious, long-term project to make all systematic knowledge immediately computable by anyone. You enter your question or calculation, and Wolfram|Alpha uses its built-in algorithms and growing collection of data to compute the answer.”
In other words, the objective is to be able to produce a “report” of all the world’s digital, computable information with one concise search. At this point, they’ve got a nice working model, but the information in the engine isn’t as good as it will be in the future. For instance, when I put in “Hippopotamus and African Elephant,” I got much more info on the Hippo than the Elephant because it’s a species (silly me). I just love the visual tree created that shows the relationship between the two animals. (Click here to take a peek.)
Will Wolfram|Alpha eventually include encyclopedia-style information? There is some debate on this. Currently, it doesn’t, but this first step is a good one and I’m excited about its potential. As soon as I got the “report” concept of what this engine was trying to do, I had visions of being able to type in a question and having a full report generated – facts, figure, prevailing point of view, dissenting views, bibliographies where I could learn more, etc. This would leave me more time to think about what I was learning. I believe the general public would be excited too. For years I’ve been hearing consumers complain about how hard it is to find things online, how much time they spend looking for information, how results are often not related to what they’re searching for, etc. Information overload…managed.
At the same time, consider the work being done by Erik Hersman and the Ushahidi website, where the content of cell-phone text messages, sent during a crisis (earthquake, terror attack, etc.), are being analyzed for content so that news can be shared with the rest of the world much more quickly. I encourage you to take a few moments and listen to Erik’s speech at TED: click here to listen.
Putting the two ideas of fact and language-based computable data together, it’s easy to see how businesses will be able to use the newer engines to get a quick read on things like customer satisfaction, new product launches, and so on. And it could be done globally or locally. In short, this evolution in our access to knowledge will allow us more time to analyze information and to subsequently develop better plans, ask better, more nuanced, questions, and more. Exciting times!
