How Do Companies Name That Price? A.I. Can Help
By Robert J. Bowman | SupplyChainBrain
When it comes to determining the optimal price of a product, component or raw material, artificial intelligence can do more than come up with the answer. It can help you to understand the questions you ought to be asking.
That, at least, is how Geoff Webb, vice president of product marketing with PROS, sees the application of A.I. to the complex exercise of pricing.
A.I., of course, has the potential to affect multiple stages and functions of the supply chain. But pricing — given its dependence on a host of ever-changing factors — seems an especially fertile target for the technology.
In its purest form, A.I. represents an attempt to mirror the workings of the human brain. When being applied to business and the supply chain, however, it takes on a different form — one that’s both more and less “human” in nature.
On one hand, A.I. can generate “expert systems” that seek to replicate the knowledge of seasoned veterans in a given field. On the other, it can far surpass the ability of the brain to crunch numbers and analyze huge volumes of information.
It’s that latter quality that applies to data-heavy disciplines such as pricing. The relevant factors that go into determining the right price of an item are far too complex and voluminous for a human to effectively consider. Webb sums up the role of A.I. in this way: “It provides an extra capacity to manage and extract insights and intelligence from information.”
Machine learning, an aspect of A.I. that improves an automated system’s skills over time and experience, is a key part of the puzzle. While theoretically replicating the learning curve of humans, it can also help managers to deliver faster and better outcomes, says Webb.
In its current stage of development, A.I. serves as a “co-worker” that augments human decision-making, he says. Among its greatest strengths is the ability to monitor variations in raw-material costs, driven by economic cycles, changes in supplier stability, and tax policies, to name but a few relevant factors.
The analysis ripples through the many stages of a supply chain, each of which can have a major impact on the final price of a product. At the same time, an A.I.-driven system can help to sort signal from noise: it can determine the point at which the model is in danger of being swamped by too much data, not all of it relevant. “Certain elements are highly impactful,” notes Webb, “but above a certain number you don’t get a lot of bang for your buck.”
A.I. can be of particular value in sussing out the interactions between different types of inputs — “places where humans may not realize that these types of variables are grouped together,” says Webb. It’s all about pattern matching, one of the abilities that is central to the workings of the human brain, but on a much greater scale than a person can handle.
Global supply chains might be dealing with billions of transactions on a monthly basis. A.I. can identify relationships and commonalities among variables that are invisible to even the most highly skilled human. “The result,” says Webb, “is much greater richness and breadth of data.”
All of this happens with a speed that can’t be matched by a human-driven operation. Such a quality is of increasing importance to organizations today, given the rapid pace of global commerce. “Managers don’t have time to grind through the numbers to come up with optimum pricing,” says Webb. “It has to happen much faster than it ever did before.”
Can the A.I.-driven model become so fast and complex that it can’t be understood by its human users? That’s a possibility. One could envision a “black box” that spits out answers that managers would have to take on faith, having no visibility to the calculations that generated them.
“Much of what we work to do is establish credibility and trust that A.I. can actually make the right decisions,” admits Webb. At the same time, the results tend to speak for themselves, in the form of pricing that improves the bottom line.
Finally, there’s the question of whether a human-run organization can keep pace with the changes dictated by an A.I. pricing model. “If you can’t execute on it,” says Webb, “it’s nothing more than a really expensive set of paper reports.”
The challenge then becomes to put into place a system for acting on the model’s pricing recommendations in a timely manner. Companies might be juggling hundreds of price lists, each of which needs to be kept current. Just how frequently that needs to take place varies among organizations; Webb has seen some companies update pricing every six months, while others do it twice a day. (In the latter case, the notion of a fixed price list becomes essentially meaningless.)
Webb sees the application of A.I. to pricing as continuously evolving, as businesses strive to meet customer demands and keep pace with the pricing strategies of competitors. The goal is to be able to price a specific transaction based on conditions of the moment, and deliver it quickly. In such cases, says Webb, a dynamic pricing model based on A.I. can be “incredibly powerful.”