Pilot Flying J lets artificial intelligence do the decision-making and brings fuel margin accuracy to 99.99%
Pilot Flying J supplies more than 7 billion gallons of fuel per year through its network of more than 800 retail and fueling locations that offer various products and amenities to make road travel easier. “Around 80% of what we sell is a commodity—diesel and gasoline—that you can buy almost anywhere,” said David Clothier, Pilot’s Vice President of Finance, Treasurer & Controller.
Pilot’s fuel margins are significant as they drive much of the bottom line and profit. If there is an error in fuel pricing at a location, they can either lose customers because the price is too high or lose revenue through low pricing for customers who would have visited anyway. With a 25-person finance team spending hundreds of hours during the financial close reviewing over 25+ P&L line items across 800 locations to make sure there are no errors in fuel margins, David knew there had to be a better way.
Pilot Flying J partnered with Infor’s Applied Innovation group, which is tasked with bringing artificial intelligence to life in real-world scenarios, to use Infor Coleman Machine Learning (ML) to automate this manual, time-consuming, and potentially error-prone fuel margin anomaly checking process in the finance department.
“Our CEO is awesome. He’s built this company from scratch and is very hands-on. He likes to be out in the stores and has the uncanny ability to ask a store manager for their financial report and spot a fuel margin error. He calls me up and asks why the margin down the street is 5 cents more. That has potential repercussions because the fuel business is a penny profit business. One penny makes an enormous amount of difference. You need to have it right; it’s got to be accurate because we are going to sell 7.5 billion gallons this year.”
David Clothier, Vice President of Finance, Treasurer & Controller, Pilot Flying J
Like all Infor CloudSuite customers, Pilot already owned the innovation technology platform in which data, analytics, and AI and ML services are pre-integrated. Pilot ran Coleman machine learning on 36 months of historical data residing in the Infor Data Lake to highlight fuel margin anomalies- - any line item that calculates a large difference in the machine-generated margin to the actual margin.
“Infor brings to the table a suite of products that we think is as powerful as any ERP on the market, the ability to get your business problems solved is unparalleled. But not everybody has an artificial intelligence engine that sits atop that, and not everybody has an integrated analytics platform that can visualize, analyze, and report on data. So, when you bring all that together, that’s the power. Nobody else does that, and what’s more, Infor is genuinely interested in solving our problems,” says Clothier.
In just under 90 days, Pilot’s AI application was tested, validated, and operationalized. Because the Infor OS cloud technology platform is pre-integrated, data is automatically updated, calculated, and anomalies presented to the finance team in Infor Birst dashboards. Instead of the finance team tediously looking at thousands of lines within a report to check for fuel margin errors, the dashboard visually presents the few anomalies that need human investigation and resolution. Does it make sense versus budget? Does it make sense versus last year? Is the trend okay? The finance team can answer all these questions quickly and accurately with a dashboard to drill and ad-hoc capabilities accessible at their fingertips.
Saving time not only saves Pilot money in resources, but the faster fuel margins are corrected, the better. The business results expected are phenomenal, both financially and operationally:
- With up to 99.99% accuracy in fuel margins, better pricing and supply decisions can easily add many basis points to the fuel margin, where an increase of 1/100th of one penny can mean $750,000 a year
- With the automated machine checking of fuel margin errors, the equivalent of two FTEs can be redeployed to other finance work
- A real artificial intelligence application deployed in 90 days
- 1,000% ROI attributed to FTE savings and $0.001 per gallon increase in revenue without procuring additional hardware, software, or specialized skills
- 0.5% improvement in P&L accuracy
- Improved employee satisfaction because the finance team are passionate about getting the numbers right
David Clothier says that “We didn’t have to purchase any additional software to create an innovative service to deliver to our finance team. We leveraged the technology we already owned to solve a business problem. There’s minimal incremental cost, so the ROI is fantastic.”
After validating AI’s considerable value to the business, Coleman ML models will be leveraged for the next major AI project. Clothier’s vision is to use AI to optimize Pilot’s workforce scheduling. With detailed time-stamped data at the item and store level, Pilot expects to predict the next 13 weeks of sales revenue down to the department level. These predictions will then be fed into the workforce management system to optimize scheduling to ensure the right employee works at the right time of day.
“Retail is constantly changing, and that’s why you’ve got to be constantly on your toes. We see electric vehicles; there will be autonomous vehicles, even trucks powered by hydrogen in the future. AI-powered insights will be key to improving the customer experience, being accurate, and ensuring we are as competitive as possible. We are ready for the future; we pride ourselves on that” summarizes Clothier.