Inflation is the highest it's been in decades in the U.S., according to the U.S. government.
The consumer price index increased 0.8% in November, pushing the rate of U.S. inflation to a nearly 40-year high. Consumers are feeling the effects of the increase across all sectors, and many retailers have responded to rising gas prices and supply chain issues by raising prices.
However, some believe retailers can provide consumers with a win-win situation through AI tools.
Matthew Pavich is senior director of retail innovation at Revionics, an Aptos company. The vendor provides retailers AI tools that help optimize prices across all items. In this Q&A, Pavich explains that the merchants that come out ahead of inflation are using predictive analysis to determine which prices to raise and which to leave alone.
How has AI affected inflation, and how has inflation affected AI?
Matthew Pavich: Inflation comes first. There are problems that lead to cost increase, whether it's production problems, drought for some of those commodities, products for groceries or supply chain labor.
AI enables retailers to surgically balance where they're going to take price increases versus price decreases. Because they're using AI, they're able to understand consumer demand patterns that are different on different products, in different categories, in different locations.
AI has been effective for retailers. A lot of our retailers are using the science to determine "OK, these products are most important to our consumers, we will not take a price increase." In fact, in some cases, we'll take price decreases on them, even though our costs are going up. We'll balance it out using algorithms, using optimization across the product portfolio elsewhere. And the same goes for different markets. It ends up being a win-win situation, where consumers are getting better prices on the products they care most about and the retailers are winning because the science is enabling them to grow both profits and then share on those items that they're not taking prices up on.
How can companies create AI tools that can account for price change and inflation?
Matthew PavichSenior Director of Retail innovation, Revionics
Pavich: A real AI solution is one that continues to learn and grow as more things happen and gets smarter. Any other form of AI -- even outside of the pricing realm -- is like a robot that gets smarter as it gets more programmed.
In the case of pricing, what it truly means is, every single time you take a price change, something will happen. And the system gets smarter because of that.
If you take a product and take the price up 10% something will happen -- demand will shift, other products will sell more or less because of that change. The more pricing moves, the smarter the AI tool gets. Then it gets even better at predicting the next move.
If the tool is not learning … then it's not truly a learning model. It's just kind of a basic math solution, which some of the more entry-level pricing providers provide.
How do retailers account for times when using AI goes wrong?
Pavich: Forecasting is hard. There's a reason nobody gets the March Madness bracket correct. So yes, you will have some things that happen where you will make a prediction, and it might not be perfect.
The beauty of AI is it learns from that and says "Wait, you know this pricing move didn't produce the exact forecasted outcome that we predicted," and it gets smarter and the next time it will be closer from a forecast perspective.
The thing that is great, though, about the AI tool is it takes the data that you have and always comes up with the best forecast possible based on the information available. Sometimes, there's just products that don't have enough information.
It's important to have good data quality. Ultimately, we'll make the best prediction based on the available data that we have. Nine times out of 10, it's going to always be better than what a human could predict.
Can an AI tool help retailers get ahead of inflation?
Pavich: It depends.
Retail does have the advantage of seeing what's covered in a lot of ways. What's nice about predictive analytics is you can technically get ahead of it from a pricing perspective. You can model out what would happen if vendor A gives me a 20% cost increase on all my products? How do I proactively set my prices in the market to prepare for that, and to set the market for that in a way where, again, I'm taking prices up on the products that are more inelastic, that customers are willing to bear that price increase on. And I'm taking prices maybe down or in a different direction or not changing them on the ones that are most important to customers. So, you can get in front of it in some ways.
What keeps some companies from using retail AI tools to help with inflation?
Pavich: Psychology studies and consumer behavior shows people are inherently risk averse. There're tons of studies out there, whether it's Kahneman and Tversky, or whichever ones you want to look at, but it's a proven fact that people are comfortable with what they do.
If you think about retail, retail is very much an industry where people do their own forecasting. They very much historically like to do what they did the year before, because it's easy to forecast. If I run Coke at $2 off [in the] first week of September this year, and I run Coke $2 off [in the] first week of September next year, that's easy to forecast. You know what will happen. But that doesn't mean it's the right price. It doesn't mean it's the correct thing to do. But it's easier, it's more risk averse, and a lot of merchants are very successful in their career and have controlled pricing for years and feel like they're good at what they do. They don't want some computer or some company telling them how to run their business. That's just human nature.
What are some of the challenges of using retail AI tools to solve or combat inflation?
Pavich: At the end of the day with inflation, I think most retailers want to wait until they absolutely must. Usually, they will wait till they get the cost increases. They do have those predictive analytics I looked at. But I think one of the things that also is important to them is a competitive monitoring to figure out what their competitors are doing and understanding who moves first and who follows who.
So that's another area where, for instance, we offer those types of analytics so we can see which retailers are following each other, and which ones are not following each other? If I were to take my price down 10% in this category and retailer A follows me, do they match me? Do they beat me by 5%? Do they beat me by 10%?
Once you can start figuring out and reverse engineering your competitive strategy with the data you have and the competitive intelligence you have, that can be powerful -- especially in an inflationary period. You can start understanding who moves first, who waits for somebody to move, who when they wait to move decides to beat the competition or decides to go higher than the competition. So, you can really lay out all the variables in what's happening in the market to find out what that optimal solution is.
Editor's note: This interview was edited for length and clarity.