Person 1: “You are my girlfriend.”
Person 2: “I’m not your girlfriend.”
Person 1: “Yes, you are!”
Person 2: “I’m casually dating a number of people.”
Person 1: “But you spend twice as much time with me. I’m a clear outlier.”
Person 1: “I must be your statistical significant other.”
The term “minimizing regret” in scientific circles can be loosely defined as minimizing patient loss during clinical trials. A scary proposition to bring up in a marketing-related blog, but a reality of scientific experimentation. This is also a key term if you’ve ever been to Las Vegas and lost your money on slot machines. Not how much can I win, but how long can I play before losing? Regret theory is a fascinating side of human behavior.
Most believe marketing’s key principle is to spend the least amount you can to derive the maximum “impact” for your company in the shortest time possible, right? Impact is a key term here. It has wildly different meanings to each business, but for the sake of this argument, let’s say impact is revenue increase.
Most look at regret theory from a customer perspective and somewhat of a long-tail view. Are consumers happy with their purchase, will they buy again, will they have buyer’s remorse when the find out they could have purchased this item last week cheaper or from another provider at a lower price? All those are keys to customer satisfaction.
I tend to look at regret in terms of wasted marketing spend. The ad industry will say 70% of ads served are wasted media — and thus the emergence of RTB and other more automated ways to optimize spend with the least amount of regret.
If you look through this lens, you will think differently about speed and efficiency. How can I act like a scientist and run experiments to prove or disprove my hypothesis (e.g., increased discounting increases average order value, yet reduces Lifetime value as the cost of increased discounting early in lifecycle can create discount habits that erode profit margin over time.)
The problem with having a scientific mind in an ad-hoc marketing and advertising world is, most don’t have the patience to sit and wait for an experiment to finish before moving on the budget. In most cases, there are many if not hundreds of experiments running all the time, being impacted by environmental variances out of your control.
What most don’t realize is that machine learning isn’t their only answer. There is a hierarchy to what we call marketing analytics. You have an algorithm, which is a predetermined set of rules for computational steps that produce a computational effect — basically “if/then” rules.
The challenge with algorithms is they aren’t typically sensitive enough to context. I’ll give you an example. No model can factor in things like a customer’s emotional or physical state — like, if he or she slept well or not at all the night before.
Sure, we can correlate weather patterns to potential emotional states by region. The first snow day of the year may be a totally different factor than the 30th day of 100 degree heat in Texas.
Algorithms can tend to stifle customers’ emotional responses to marketing offers over time. By nature, “if/then” rules typically imply a calculus (if customers are 35-55 and just paid 20% down on a mortgage, then they should purchase a home equity line of credit).
None of this includes a human element that incorporates varying emotional or physical states that may require some human interpretation. Yet the antagonist will say, there’s no way to scale human interpretation so pattern-based marketing should take over.
One thing you should not forget this holiday season is the value of unpredictability. Yes, you may violate the first principle of this article (minimize regret), but businesses value unpredictability and risk-taking in marketing.
While our goal is to move as fast as we can and minimize regret, if you can program machine-learning to focus on a percentage of your operations, it will free you up for the randomness that is the most rewarding and potentially most effective areas of marketing.
Originally Posted LinkedIn /Media Post 2016