One of the hardest questions for anyone who makes or sells products to answer is: “What’s the right price for my product?” After all, the price is probably the one thing that most influences how successful and profitable the product is going to be. Set the price too high, and sales will be so low that fixed costs cannot be recouped. Set the price too low, and the product will sell like hot cakes, but without leaving any profit in the seller's pockets.
What makes pricing so fascinating is that it’s the only one of the four strategic components of the marketing mix (the others being product, promotion and placement) that can be changed at very short notice. Open any microeconomics textbook, and you will find two key statements: “The price is determined by the customer's willingness to pay,” and “the customer's willingness to pay can be described in a price-demand curve.” Theoretically, all it takes is choosing the right point on the price-demand curve to determine the ideal price with respect to the strategic goals. But then, as any pricing practitioner will tell you, actually determining the customer's willingness to pay, let alone mapping it on a curve, is downright impossible.
This is what we discovered years ago when one of our long-standing customers asked us to help with implementing dynamic pricing in their online shop. Surveying the market, we quickly saw that the underlying problem – that actual price sensitivity is hard to measure wasn’t properly addressed in a scalable way by any of the available vendors. In lieu of a true solution, we saw workaround approaches to dynamic pricing including:
- Price-rule engines that allow the automated application and enforcement of pricing “best practices” without actually validating how good these “best practices” are (surprise surprise, they’re not that good).
- Price robots that trawl competitors’ websites for information and automatically make changes when they find lower prices. While this approach is market-driven, it’s supply-based, not demand-based and assumes (incorrectly) that customers and shops are completely interchangeable.
- Targeted pricing frameworks that allow the creation of different pricing tiers, for instance higher prices for returning customers and loss-leading prices for new customers (or vice versa). This is all very well, but it doesn’t answer the underlying question – “What is the customer (logged in or not) willing to pay?”
So Blue Yonder's data scientists approached the problem by taking a science-based approach and running an experiment. Not just a small, try-it-at-home experiment, but a big, bold, systematic experiment to measure the impact of price changes on demand. After a couple of weeks and almost a million price changes, the experiment confirmed a couple of things:
- The price-demand relationship exists, and given enough data it can be learned by an algorithm. Understanding the impact of price on demand requires accounting for all other influencing factors such as day of the week, product presentation in the online store, ongoing promotions, product category and description.
- Once the price elasticity of demand has been measured, it can be combined with multiple optimization functions to maximize revenue growth, increase margins and profits or achieve a combination of both.
In the end, the scientific approach to price optimization proved to be a huge success, and even during the experimental phase it paid dividends with increased revenue and margins. Our professional services team continued to work with this and other customers to roll out price optimization to more and more stores and inventories. At the same time our product development team took on the challenge of bringing scientifically sound price optimization to more e-commerce retailer, focusing on:
- Standardized APIs, allowing existing online-store software to be integrated easily
- More robust algorithms for revenue and margin optimization
- A scalable multi-tenant runtime, so that new customers can be brought on board in a short time and benefit from increased revenue and margins within a month of go-live
I'm proud to announce the general availability of our newest predictive application: Forward Pricing. Forward Pricing is a dynamic price-optimization solution for e-commerce retailers that automatically and systematically performs price experiments, measures how price changes affect demand changes and uses that knowledge to determine the optimal price for each product, in accordance with the company's overall strategy.
Today, Blue Yonder's price optimization is managing many hundreds of thousands of products, has performed tens of millions of price changes and has realized tens of millions in additional revenue. By tomorrow, you will be able to benefit, too.