Deductive reasoning in the age of Big Data and Pervasive Marketing Technology
One of the more challenging aspects for today’s marketers is the analysis of large data sets and the use of sound deductive reasoning…
…the ability to reach good decisions based on the data in front of them.
Many people, including marketers, are capable technicians. They can build statistical models, write code, setup and manage automated processes in the tools of the trade, but doing analysis, interpreting data and drawing useful and correct conclusions is elusive. Too often, marketers, executives and statisticians see what they want to see and find the data to support their own emotional bias or wishful thinking.
For the fun of it, and for the controversy that surrounds it, I’d like to use global warming for illustration purposes, before returning to marketers and the dilemma they face in a massive sea of data.
Perhaps we can all agree we want to be good stewards of the earth and our natural resources. However, we may not agree on the conclusions being derived from the evidence presented. Let me offer a few questions that may challenge today’s accepted norm – in other words, let’s play what if and challenge the foregone conclusion on global warming that…“everyone knows the science is factual and only idiots will think otherwise.”
Question #1: What did man do to cause the end of the last ice age? Clearly if mankind is the cause of global warming, then the obvious relevant question is, how did we end the last ice age? As reference, some authorities suggest the last ice age ended about 11,500 years ago.
Source: http://www.bbc.co.uk/science/earth/earth_timeline/quaternary_ice_age
Question #2: What did mankind do to cause the earliest recognized ice age? If we are the cause of global warming, then the reverse conclusion has merit. In other words, on the basis of what action(s), or the absence thereof, did we caused the earliest ice age? The earliest ice age started about 2.6 million years ago.
Source: http://www.bbc.co.uk/science/earth/earth_timeline/quaternary_ice_age
Question #3: When the earth rotates around the sun, how does it warm the sun? Oh it doesn’t? Okay, then what does the earth do to cause the sun to warm mother earth? Ah, perhaps the earth generates all of its own heat and the sun has no influence on the warming of our little globe. San Diego State Geological Department identifies extraterrestrial impacts, gravitational contraction and radioactive decay as the three main sources of Earth's internal heat.
https://www.reference.com/science/three-main-sources-earth-s-internal-heat-7a76e0ef97c6ad35
In contrast the Ohio State University points to the Sun as the primary energy source of earth. They further state, “Gradual changes in Earth’s rotation and orbit around the Sun change the intensity of sunlight received in our planet’s polar and equatorial regions. For at least the last 1 million years, these changes occurred in 100,000-year cycles that produced ice ages and the shorter warm periods between them.”
Clearly this thing called “global warming” or “climate change” as we now prefer to call it… is a complicated topic and requires some more study.
Question #4: What is my sample size and is it statistically valid? When we talk about climate change and authorities make statements like the following, “The World is Warming. It’s Because of Us. We’re Sure” (National Geographic, April 2017) and reference data that shows global temperatures for the last 116 years as the proof point… is this a statistically valid sample size against 2.6 million years, or 100,000 year cycles, or even against the natural cycles of our earth, which has existed for 4.55 billion years?
Source: https://www.scientificamerican.com/article/how-science-figured-out-the-age-of-the-earth/
Question #5: Where is my control group? In marketing we are often taught about establishing control groups in order to minimize the confusion that large numbers of variables might cause in drawing predictable, reliable conclusions. The same is true in scientific discovery, using the scientific method and control groups.
http://www.sciencebuddies.org/science-fair-projects/project_scientific_method.shtml
https://www.thoughtco.com/what-is-a-control-group-606107 definition by by Anne Marie Helmenstine, Ph.D.
So… where is my control group for planet earth? Oh, we don’t have one. Planet earth is the only planet of its kind where science can point to a large population of living humans.
Okay, so now that I’ve set the stage using global warming for illustration, let’s return to today’s marketing realm.
Question #1: What did you your customer do to end the latest drought in your sales production? The most obvious answer is they bought your product. But why? Why did they suddenly start buying your product again?
Question #2: What did your customer do to launch your company? Oh, they didn’t? You started your company… perhaps even you started it to fill a customer need and they were happy to pay you to satisfy that need.
Question #3: When your company hovers and rotates around your customers – you know, capturing a 360 degree view of customer behavior, what do you do with all that data to create new products and generate more revenue and profits?
To put this into perspective: 1.2 million products, $15 billion in revenue; 15,000 employees, 100 million customer interactions monthly via telephone, mobile device, web site, email, text, social posts, etc. Geographic data points, demographics, firmagraphics, transactional details, including real-time, variable content, personalized to every individual at the right time, in the right place with the right product. Each interaction, unique and personalized to the user…. Get the picture? There’s a lot of data in here and it’s only going to get worse with the Internet of Things showing up.
Question #4: What is your sample size? Is your sample size a single interaction, a single purchase, a single conversation? Or a billion customer interactions? And is it statistically significant, valid and representative for the decision you’re trying to make?
Question #5: Where is your control group for your 360 degree customer view? Oh, you don’t have one? Better yet, what is the definition of your control group and how will you eliminate thousands or millions of variables in order to isolate cause and effect and produce predictable performance?
As you can see, today’s marketers have a bit of a challenge when it comes to doing analysis and reaching sound conclusions.
Early in my career, in the magazine publishing industry, I enjoyed the benefits of a stable product with stable pricing and a stable control piece that lasted multiple years. It was tough to beat! When I moved into catalog retailing of technology products, I suddenly had 150,000 products, with declining prices and rapid product obsolescence. The rules of the game had dramatically changed and my control group practices, data analysis and assumptions likewise had to change. In a nutshell, I needed new perspectives and new skills.
Today’s marketing leaders must have or must develop excellent skills in analysis, data interpretation and deductive reasoning. If they don’t then like so many others in their companies, they will simply see what they want to see and look for data that supports their own emotional bias and wishful thinking.
We need marketing leaders that have wisdom, discernment and backbone to use sound deductive reasoning. If the prevailing opinions are overwhelmingly popular, but lack the support of basic, solid analysis (e.g. no control group, non-existent or miniscule sample sizes, short windows of time and a million variables), then stand up and admit to your CEO exactly what you’re actually doing – taking a best guess to support your own wishful thinking.
As for climate change – that’s an easy one– it happens every day, every winter, spring, summer and fall.