Behavioral Economics Powered By Machine Learning And AI Techniques

A true path to optimized pricing

There is a growing focus among pricing practitioners to use price science to drive pricing decisions. It’s a fabulous trend with many recent innovations in the use of advanced techniques such as machine learning and AI. It is important however to consider behavioral economics because our economic decision making is often shockingly irrational. Naturally, we are not referring to your personal decisions, dear reader, because you are always rational, right?

Consider a recent TED presentation by Dan Ariely, Professor of Psychology and Behavioral Economics at Duke University, Are We In Control of Our Own Decisions? Professor Ariely shows a number of highly engaging examples of irrational decision making that includes an intriguing pricing study.

Professor Ariely studied three subscription tiers for a well-known periodical. The tiers were as follows:

1. On-line only option $59
2. Print option $125
3. Print and on-line option $125

Is this pricing strategy a mistake? Perhaps it was, but further analysis proved interesting. The study first showed expected buying behavior:

Clearly the print only option had zero economic utility and was thus meaningless. But, what do you think happened when this option was eliminated?

Wow, what just happened? Behavioral economics suggests that that our decision making is in part predicated on immediate reference points. The study indicates that offer three looked more desirable when compared to a weak alternative (Offer 2). Apparently, economic preferences and associated purchase decisions are more malleable than we as individuals perceive. Are you making the decision or was the decision largely made for you through a cleverly constructed offer?

This brings us to today’s point. How can pricing practitioners best combine advanced predictive analytics and behavioral economics? Advanced algorithms do a very good job of using pricing history to predict the most likely outcome of a price change, but within the confines of what was manifest in the training data set. And AI is often modeled after the assumption that buyers are utility maximizers. So, what happens when buyer behavior is irrational?

Incorporating behavioral economics into pricing models can be achieved by the scientific method in a process that consists of (1) domain experts conjecturing a hypothesized strategy, (2) surveys conducted to get market feedback on the hypothesis, (3) price experiments (e.g., A/B testing) executed in a representative market, and (4) learnings used in training data for ML algorithms and AI model definitions. Repeating the process then becomes a positive cycle of learning, anticipating buyer behavior and optimizing strategies.

Another perspective on behavioral economics is provided by Tom Brzezinski in a Vistaar webinar: “The Psychology of Pricing: Is a Penny Just a Penny?” He examines how customers perceive value and pricing. Tom has over 30 years of pricing experience, and he relates that experience to key topics focused on the question—what drives one customer to accept a price increase and another to buy your competition? Additional topics include:

• How does your brand influence a customer’s willingness to pay?
• What is the relation between everyday price and promoted prices?
• What effect will a new higher-end product have on spending?
• What is the ‘anchoring effect’ and how does it work?

Vistaar is a huge advocate of generating value through price science. With today’s pricing spotlight on topics such as machine learning and AI, we think it’s important to have a balanced perspective that includes behavioral economics.

This article was co-authored by:
Paul Greifenberger, MBA, The University of Michigan, Ross School of Management
David Glenn, PhD, The University of Michigan, Operations Research