To better price strategically, Spinx is applying artificial intelligence (AI) and business intelligence (BI) capabilities in PriceAdvantage’s fuel-price management software. Its goal: know its own fuel price and in-store dynamics and predict the moves of its competition.
The Predictive Model tool in PriceAdvantage uses AI and machine learning to analyse historical data to predict how a retailer’s competition will react to its fuel price changes, as well as the effects on volume and margin.
“It almost gives you an insider tip into their business based on the data we have,” said Nick Perry Director of fuel. For example, the last 10 times that fuel pricing conditions were this way, the competition reacted this way. “In pricing, if you know they’re going to do that, you can get out there and sharpen the pencil, take a few gallons prior to that happening knowing you’re going to get a restoration of the market,” he said.
The tech can also flag if a competitor’s pricing patterns change; from there, Spinx can experiment, watch and analysis its competitor’s reaction to price changes to see if it is a fluke or part of a new approach.
“We have a competitor that is in all three of our markets that is acting abnormally,” he said. “They presented a new pattern to us. The first time we saw it was an oddball, so we red-flagged it; let’s pay attention and watch it. We watched it over the next month or so. Now we are going to validate our confirmation by throwing wrenches in there.”
Spinx hypothesized that the competitor had hired a new fuel director who may be following different fuel pricing rules. After the AI functionality flagged the new pricing behaviour, Spinx examined whether it was specific to one site or more and watched the trend. “It’s abnormal to us—but it might be the new normal,” Perry said.
“AI machine learning is critical for empowering and enabling these rapid decisions to be made, but it’s never going to replace that human element,” said John Keller, division director for PriceAdvantage, a division of Skyline Products based in Colorado Springs, Colo. Perry of Spinx can adjust his pricing strategies for a period of time, analyse how the competitor reacts and test out theories. If they prove to be true, the software’s machine learning reorients in real time to build a new model on the competitor.
“It’s dynamic,” Keller said. “You’ve got razes-and-rebuilds, acquisitions, even simple scenarios like fuel analysts taking a vacation for a while and you have someone else in their place. All these parts are part of the equation.”
The BI and AI functionality also help Spinx better understand their pricing sweet spots, down to the individual store. “There are some stores that through trial and error and data mining, I don’t need to be the leader—I’m getting what I need,” Perry said. “I don’t need to go out there and drag the street margin down. At the end of the day it’s putting dollars in the bank.”
Perry described the AI potential for fuel pricing as “almost scary.” That’s because it requires a discipline. “Yes, the data is there, but the human is the one who told the computer to go look at this data, and you might be looking at the wrong template or data set,” he said. Or there could be a factor such as bad weather or a pipeline outage that needs to be taken into account.
Spinx has been able to re-prioritise its employees’ responsibilities with the AI and BI capabilities of the fuel pricing software. Two years ago, it had store managers stop surveying competitors. “There were 7 minutes on average that manager every morning had to spend writing down a survey and keying it in,” Perry said. “Now they spend it getting the store ready for the day.”