Skip to main content
Four Axioms for Big Data Marketing

By Josh Hemann

A few weeks ago there was a fair bit of hoopla (in certain circles at least) about how Oreo won the Super Bowl. How exactly? By having a team of marketing people posting on social media channels, in real-time as the Super Bowl played out. When power was lost at the stadium Oreo delivered a timely and funny tweet. What was striking was the ensuing discussion on what seems like a minor event: 15,000 people retweeted and it was hailed as a monumental triumph, a coup d'etat in the marketing world where big fish were paying millions for 30 second TV spots. Seriously?

I am not sure what constitutes success in these circles, but I have a feeling retweet rates are not correlated with profits. Further, I have a feeling that what is going on here is just plain narcissism: marketers captivated by the reflection of themselves in the modern day lakes of Facebook likes and Pinterest pins. (A friend who is a chef jokes that chefs create menus to impress other chefs, not customers.)

In the same vein, there is such self-serving hyperbole about what social media or Big Data or Hadoop can do for businesses that it is easy to lose sight of fundamentals. Cathy O'Neil, in her post The smell test for big data, writes about the need to apply common sense to some of the claims running wild these days:

If I can imagine it happening in real life, between people, then I can imagine it happening in a social medium. If it doesn’t happen in real life, it doesn’t magically appear on the internet.


I have been thinking about how to stay grounded at times like these, when there is so much change in:

  • The state of Retail, the retailification of non-retail businesses (yes, retailification is a word. I think), and Marketing in general
  • Mathematics for characterizing and learning from huge volumes of disparate data
  • Software implementations of said mathematics

and I am reminded of one of the first things I learned when studying probability: its axioms. In mathematics, axioms are fundamental statements upon which more advanced statements are developed. But the axioms themselves are radical in the true sense of the word. They are not proven, they simply are. Here are some axioms in the world of Big Data Marketing that help me keep my sanity and have a fighting chance of doing meaningful work...

    1. Humans are social animals. Once basic food, shelter and security needs are met, we want to spend time socializing. If you want to skip some of the research backing up this statement, spend an hour watching the movie Happy. To hear many marketers talk though, you'd think people have nothing better to do than consume, think about consuming, and talk about consuming. The Onion wrote a satirical piece about this topic almost 10 years ago and the skewed reality has only gotten more skewed. It's a healthy reminder that my mental and emotional energy should be spent on people, not brands and material goods. This brings me to... 
    2. Consumers are loyal to brands and businesses in so far as those brands and businesses make inefficient markets more efficient. Here, I am using the phrase “inefficient market” not in the academic sense, but in a more colloquial sense. Strong brands and businesses tend to do one or more of the following really well (all backed by data and analytics):
      1. They provide stuff people want, make it easy to discover, and easy to get
      2. They enable markets by connecting buyers and sellers, consumers and producers, in a new setting
      3. They enable mental models of how to assign value to products and services
    3. Customer-centricity is about more than using data to create crafty marketing. While customer-level data is often generated through marketing-related touch points, such data should be used throughout a business’s decisions and processes. 
    4. Tools impose structure. Be mindful of the tools you choose to use to tackle analytics, for they never stand in isolation. All tools come with a culture around them and operate within an ecosystem of other tools. Poor choices in tools will bias what types of questions you ask as well as affect your ability to communicate information.

    Putting these axioms to use

    Here are some examples of how I use these axioms…

    • Marketing analytics – I increasingly find myself working with retailers who want to test complex offer strategies because they think that is what Big Data and analytics enable. For example, someone wants to test how to best target customers with Buy X and get Y dollars off of Z type offers, as well as test different flavors of these offers. The first two axioms come to play here: just because you can build a predictive model to support such targeting, why would you? The whole premise is flawed. Analytics and data should be used to make purchasing simpler and faster, not in forcing people to go through mental gyrations when buying commodities. This belief is axiomatic: I am not going to spend time debating it or testing it. Instead, I am going to build subsequent debates and testing from this position.
    • Pulling big levers – Too many of the retail analytic discussions I find myself in focus solely on how to target people with the right ads and coupons. Relevant communications are the cornerstone of modern marketing, but they are not sufficient for an organization striving to be customer-centric. Big Data and the customer lens should inform merchandizing, real estate, customer service, etc., as these are often bigger levers a company can pull in the attempt to create repeat, loyal customers.
    •  Working backwards – I like to start with a business problem and then assess what data are available to tackle the problem. Given the data and problem, only then do I start thinking about software.  You might be thinking, “yeah, so what?” The nuance here is that I routinely see analytic practitioners working the other way, even if unintentionally. They go through their mental Rolodex of approaches easily supported in the tool(s) they use and everything about the solution is constrained from there. But the reality of working with Big Data and complex business problems is that polyglotism is now a requirement. You need to accept the fact that in many situations you’ll need a mix of programming languages, databases, and software tools to get the job done. So, focus on the problem first and grab the right tools out of the toolbox later. Implicit here is your toolbox having a lot of tools in it. If your toolbox is limited, you’ll likely find your approaches limited from the start, whether you realize it or not.    

     These are some of the axioms I practice remembering. What are yours?

    related posts