Podcast Guest: Robert Phillips
In this episode, we spoke with Robert Phillips about pricing - specifically, dynamic pricing. He is currently an independent consultant and author specializing in the application of data science to business.
Previously, he was CEO of Decision Focus Incorporated which pioneered the development of dynamic pricing and revenue management to airlines, hotels, rental cars and many other industries and the Founder of Nomis Solutions that provides dynamic pricing software to financial service companies.
More recently, he has served as Head of Marketplace Data Science for Uber and Pricing Data Science at Amazon. He has taught courses at the MBA, Masters and PhD levels at Stanford University and Columbia University Business Schools and written books on Pricing and Revenue Management.
While we discussed analytics-led transformation related to pricing, the lessons from this case can apply to many situations which require organization or department-wide adoption of new data-driven technologies, including AI
We covered topics including -
How applying analytics to pricing was adopted by different industries
Organizational resistance to change and why management support is so important
How data-driven transformations are different today
What to do and what not to do when pursuing analytics-driven change
Why customer acceptance matters
Highlights from the Episode
What is dynamic pricing?
Oxford University Press defines “dynamic pricing” as “The practice of varying the price for a product or service to reflect changing market conditions; in particular, the charging of a higher price at a time of greater demand.” Dynamic pricing is also about lowering prices.
The airlines industry as the driver of early dynamic pricing
American Airlines is credited with coming up with the basic idea behind dynamic pricing after the government deregulated the airline industry in 1978 and allowed complete flexibility in scheduling, routing and pricing.
The airline initially offered low fares that matched their low cost competitors but limited their sale to reserve some seats on each flight for higher-paying customers who booked later. Analytics were used to forecast demand for different types of customers and determine how many seats to save for late-booking high-paying customers.
Overcoming resistance to change
Adapting to change becomes easier when there is internal and external validation of the efforts. For example, in the airlines, those that implemented dynamic pricing were outperforming those that did not (internal). Soon, there were external analysts asking airlines about their revenue management capabilities on earnings call. Clearly, it had become a critical business capability.
Resistance to change is stronger when there is a long-existing way of doing business. It then requires extra effort to implement systems successfully in the face of this resistance.
Pricing analytics at Uber and Amazon
Most tech companies have analytics in their DNA. Thus, there is no ‘transformation’ required. The expertise required to implement these systems does not come from industry knowledge as much as from analytics knowledge. That said, it is important for the data scientists to spend time learning the business in order to be successful.
“When Amazon started selling large numbers of books, they hired data scientists and software engineers who could develop and implement the algorithms that could do it efficiently and at scale. At Uber and Lyft – they didn’t hire former taxi-drivers, they hired data scientists to build their routing and pricing functions.”
How to develop dynamic pricing models
“The most successful dynamic pricing projects are incremental to some degree – prove the concept first then continue to improve – as opposed to trying to “boil the ocean” by committing to build the ultimate system right off the bat”.
The data needed depends on the industry setting
For the first airline systems, the critical data was historical sales and flight capacity. Historical sales are used to determine whether or not a flight would reach capacity and then raise or lower the price accordingly.
At Uber, the critical factor is the supply/demand balance at any time in a city – if the current and immediately predicted demand exceeds the number of drivers available you may need to raise the price (and vice versa).
At Amazon, the primary factor in pricing is, in most cases, the competitive price – Amazon accesses the prices of both physical and on-line competitor’s stores in order to set their price.
Potential pitfalls in adopting new data-driven approaches
The first thing is to make sure that you have full management support before proceeding – especially if you are introducing a ‘new’ analytics approach to an organization. If you are displacing or supplementing an existing human-driven process, you can anticipate plenty of pushback. As Robert Phillips says, “If you don’t have full management support at the appropriate level, there is the risk that the system will be rejected like a bad kidney”.
A second pitfall is not understanding the business sufficiently well before proceeding. “When starting in a new industry, I would insist that the team spend at least a week in a deep dive into the business so we could gain sufficient domain knowledge”.
A major pitfall is underestimating the time that you will need to spend on data cleansing and reconciliation. “This is fairly common, particularly when a new analytics approach is being introduced to an industry or company. In many cases, you may be looking at data at a level of detail that has not been previously considered, and you will uncover lots of unsuspected glitches”.
Customer Acceptance
Customers tend to be skeptical of any change in the way that prices are set. In the case of dynamic pricing, they get upset when they see it as a way for the business to extract more money from them through higher prices. In reality, in most dynamic pricing implementations, prices are lowered as often as they are raised. Customer education can probably help convey the right message.
New industries need to be careful about introducing dynamic prices. They should be sensitive to potential consumer concerns, but “shouldn’t let initial consumer resistance dissuade them”.
References (books)
Book - Pricing and Revenue Optimization by Robert L. Phillips
The Strategy and Tactics of Pricing: A Guide to Growing More Profitably by Muller, Nagle and Gruyaert.
Share this post