By Dr. Ted Gaubert.
Why are we paying different prices? Is it ‘price personalization’ or a form of ‘price discrimination’? The answer isn’t so simple.
The world of Artificial Intelligence (AI) dynamic pricing engines is rapidly progressing and changing the competitive landscape. This article provides an overview of a few areas that influence how an AI pricing engine decides the price to show you:
- Predicting market demand & micro-segmentation
- Accumulating asymmetric information
- Estimating ‘Willingness to Pay’
- Shaping demand
1. Predicting Market Demand and Micro-Segmentation
AI micro-segmentation uses many customer attributes and behaviors to bucket customers by estimated willingness to pay. To explain in simple terms, let’s assume we have three buckets. We segment our customers by (A) high paying customers, (B) medium paying customers and (C) low paying customers. One strategy to maximize profits would be to first sell only to the high paying customer group #A. Then any remaining seats could be sold to medium paying group #B. Lastly, any leftovers could be sold to low paying group #C .
Wait a second! We all know airline prices go up as the booking date gets closer to the departure date. Everyone knows to get a great price on an airline ticket you should book early!
So, it’s not as easy as first selling to high paying group #A, then medium paying group #B and giving leftovers to low paying group #C. In most cases the pricing and selling happens in the reverse order – Low paying group #C, then medium group #B and lastly high paying group #A.
How then are sales to the high-paying group #A maximized if selling to group #C happens first? What prevents all the seats from being sold to low paying group #C with no seats being left to sell to the high paying group #A?
Part of the answer is predicting the number of potential buyers and how much each of them is willing to pay weeks in advance. Often before most of the customers have even decided to travel!
2. Accumulating Asymmetric Information
A big part of the AI pricing game is having AI learn everything about what is happening in the market. The goal is to have better information than competitors in order to make better decisions. This information advantage is sometimes referred to as asymmetric information.
In terms of demand prediction, asymmetric information allows the AI pricing engine to achieve a more accurate demand prediction than competitors. Ultimately, this advantage results in greater confidence by the pricing engine to hold a price or move it up or down to maximize profit in response to what is happening in the market.
To see how this works, let’s take a hypothetical airline market with 3 airlines serving a destination like Porcupi, Montana. In the airline industry, Porcupi is a small town that sits in the ‘long tail’. This means it is just one of many towns and cities served where each generates only a small amount of revenue. However, like the classic example of Amazon.com, the sum of everything in the ‘long tail’ adds up to be a massive revenue number.
Suppose you are an airline operator and you know a big festival will soon take place in Porcupi. You know that significantly more people will be going to Porcupi than the number of available airplane seats across all the competitors. If you are the only airline operator who knows about the big festival, then the pricing strategy is easy. Hold the price high until all the competitors have sold all their seats. Then, travelers will have to pay a high price for a seat on the last plane into Porcupi.
However, most of the ‘long tail’ markets don’t generate enough revenue to economically justify hiring a human to monitor what’s happening in a small city and then make micro-adjustments to prices. Similar to how Amazon leverages technology to autonomously make demand prediction and pricing decisions across half a billion products, the same is happening across many other industries.
AI can learn about local events that are happening in real time on a global basis far more economically than what could ever be achieved by a group of humans. This enables a company leveraging AI to gather asymmetric information which enables better demand predictions and strategic pricing decisions.
3. Estimating Willingness To Pay
Data is constantly being collected about your customer behavior such as:
- What type of items did you look at?
- How long did you spend on each web page?
- What items did you put in your basket?
- What items did you purchase?
- What do people pay that look and behave like you?
All this data and more gets fed into an AI engine that translates your behavior into a persona and tries to predict things about you, one of them being estimating the ‘maximum price’ you are willing to pay.
Keep in mind, this doesn’t imply you will receive a ‘personalized price’ even though it is technically possible. The practice of ‘personalized pricing’ is highly debated for numerous reasons including ethical, brand, loyalty and legal concerns.
However, ‘willingness to pay’ can be used to determine how likely you will purchase an item at the current market price. This likelihood gets incorporated into demand predictions by micro-segment and, ultimately, the price. Consequently, the AI engine can control sales velocity by knowing how much to sell at what price.
4. Shaping Demand Dynamically
Predicting competitive response + demand + micro-segmentation + ‘maximum price willing to pay’ are all based on probabilities. There will always be some level of ‘error’ in the predictions. In other words, things may sell a little faster or slower than expected. AI pricing engines use dynamic demand shaping to change the shape of the demand curve by adjusting prices. This could be based on real-time inventory or any other myriad of factors. In the airline and hotel pricing world, demand shaping can be used to optimize profit and minimize overselling or underselling airline seats and hotel rooms.
Every industry is selling a product or service to a customer at some price. The topics discussed in this article are broadly applicable to a wide range of industries and business scenarios outside just dynamic pricing of airline seats and hotel rooms. Manufacturers estimate demand to know which products to produce in what quantities. Distributors manage pricing and demand forecasts to optimize inventory and distribution logistics. Marketers use demand estimations to make decisions of promotions, targeting and marketing spend. Advanced algorithmic techniques and AI engines are quietly transforming how organizations compete.
The result of all the complex algorithmic interplay determines the price we are quoted, the advertisements we are shown and the product mix we find when shopping.