The Causality Trap

IN Big Data Machine Learning — 23 April, 2015

The fact that correlation and causation are not the same is a well known fact among statisticians and data scientists. For instance, there is a distinct correlation between the number of people drowning in swimming pools in the US and the number of movies Nicolas Cage appears in. Yet not even his harshest critics would accuse Mr. Cage of causing people to drown and the hypothesis that swimming pool deaths inspire Hollywood studios to cast Mr. Cage does not align with Hollywood production timelines. 

 

 

Just being aware of a problem does not solve it. Untangling the relationship between causality and correlation has befuddled statisticians and mathematicians for decades. The resurgence of these fields in the last few years with the successful application of Big Data analytics has even led commentators to declare capitulation to the mathematical problem and victory on the application problem. In 2008, WIRED’s Chris Anderson published: The End of Theory: The Data Deluge Makes the Scientific Method Obsolete.

Since 2001 when Nassim Nicholas Taleb published his first popular book, we all know that we can be Fooled by Randomness. In our work with Blue Yonder customers we have encountered a few puzzling examples of how marketers and companies were fooled by causation.

Every marketer knows (and some like to cite) John Wanamaker’s famous quote: “Half the money I spend on advertising is wasted; the trouble is, I don’t know which half.” In fact, advances in analytics and business intelligence have seemingly enabled marketers to solve Wanamaker’s riddle.

By observing what attributes and behaviors are highly correlated with sales, marketers can identify the most valuable customers to send direct marketing material like catalogs, emails or mobile coupons to. The assumption is that by spending advertising budget on only the most valuable customers, you are not wasting money on advertising, because you can show clearly that almost every customer that you advertised to did make a purchase. Does this mean you’ve accomplished your mission? Not exactly. You have to keep in mind that some customers will also make a purchase even if you never send them any direct advertising and that some customers, who you categorize as not among the most valuable, that did receive advertising consequently made a purchase. This demonstrates that the question “Who are my most valuable customers, so I can send them advertising” is misguided and the right question would be “Who are the customers that are most likely to be influenced to purchase through advertising but would not buy otherwise?”What happens when you execute a flawed plan with maximum efficiency is what Harvard Business Review’s Roger Martin describes as The Execution Trap.

The marketer’s equivalent to the Execution Trap is the Causality Trap. Falling into the Causality Trap means using ever-advanced tools to run ever-advanced analytics to determine your most valuable customers. As you send advertising to them you are wasting more than half of your money because many of them would buy anyway. And because you not sending advertising to customers because they are not in the most valuable Segment, you are wasting the other half of your money.

Looking at the data critically reveals that fine-crafted rule books and advanced analytics lead to results that are usually no better than random results. Being in the Causality Trap means that getting better at analytics leads to becoming worse at decision making (and money making).

In the Causality Trap, your marketing activities are redundant and harmful because they obfuscate and eliminate the positive activation effects your advertising can have. By only targeting customers who would buy anyway, marketers are giving up any chance to discover the true causal effect between advertising and conversion.

Tomorrow Blue Yonder founder Prof. Dr. Michael Feindt will show how breakthroughs in Data Science enable Blue Yonder customers to escape the causality trap. You are able in the future to reduce your ad budget while keeping their yields stable. This does not mean less to double as the marketing ROI.

If you want to learn more about this topic, please write us an e-mail to info@blue-yonder.com We show how causality can be used in your Business.

Blue Yonder Blue Yonder

We enable retailers, consumer products and other companies to take a transformative approach to their core processes, automating complex decisions that deliver higher profits and customer value using artificial intelligence (AI).