From fuel margin analysis to workforce management to analyzing Twinkie sales - how Pilot Flying J plans to turn Infor Coleman AI loose on bottom-line projects
- Summary:
- For many companies, applying AI is still a hypothesis. Not for Pilot Flying J. Here's how Pilot Flying J is collaborating with Infor on a slew of use cases, including anomaly detection. Next up: fuel margin analysis, where predicting pennies has bottom-line potential.
One PR pitch that rarely works? "Our customer is about to do this." My diginomica use cases look at what customers have already done, and the lessons already learned. Our readers can't afford to linger in hypotheticals.
But every rule has its exception. I'm making one for Infor, and the AI plans of a flagship customer, Pilot Flying J.
I've written about Pilot Flying J before; they do not shy from new tech, or business model changes. They might have the word "pilot" in their name, but they do things at scale:
- They are the largest purveyor of over-the-road diesel fuel in the United States.
- Pilot Flying J is the largest travel center chain in the country with over 750 locations under the Pilot, Flying J, and Mr. Fuel brands.
- They are also the third largest franchiser of quick service restaurants in the nation.
Can AI impact fuel margin accuracy?
So why did Infor's Pilot J press release prove the exception? No, it wasn't the title (Pilot Flying J Takes Off with Infor Coleman AI & ML). It was the specifics:
Pilot Flying J, a leading network of travel centers in North America, is using Infor Coleman Artificial Intelligence (AI) and Machine Learning (ML) to help boost its fuel margin accuracy to 99.99 percent, which the company estimates could increase its revenues by $750,000 a year.
Margin precision always gets my attention:
Previously, the company’s finance team spent five days a month closing the books, and a sixth day making sure there were no errors in fuel margins. A one-penny error can have a significant impact on the business, according to David Clothier, Vice President of Finance, Treasurer and Controller for Pilot Flying J, because the company sells about seven-and-a-half billion gallons of gas per year.
Clothier sounds like a guy who knows his numbers - plus, he's been at Pilot Flying J for 29 years. Yet he knew a change was needed. Soon, Clothier was sharing his story on a video call. No surprise - Pilot Flying J is still in expansion mode. Clothier:
We continue to expand quickly. We're doing a lot of M&A... We always bring them fully on, over time, onto the Infor platform, which makes things ten times easier... Once we get you in the fam, life gets a lot easier for Pilot Travel Centers. Until we do, you're on your own set of systems, and it's a nightmare. You don't get numbers timely; you don't get people issues resolved. Nothing works.
So as an organization, that's where we're headed. We're going great guns. Like I said, the seven or eight Infor products we have all stood the test of time the last five years. We're currently actively adding to all of them. We've very active in the workforce management space; we're very active in the EAM space. I realize there's been a bit of a transition there, but as I understand it, the integrations and so forth will survive that. And the product should actually accelerate its subsequent development. [Author's note: see my colleague Derek du Preez's Why has Infor agreed to sell its EAM business to Hexagon AB?]
So how did the idea for using Coleman AI for fuel margin accuracy come about? Clothier told me that over the last few years, his team has gotten more comfortable with the idea of applying AI:
We were there when Infor announced Coleman AI. We always had in the back of our mind: 'There's some nice stuff we can do with this - I wonder when it's going to become mainstream.' It takes a while once you buy something to get it woven in. Getting it to the data and the data lake - that's not the switch of a single button. But we continued to talk to Infor.
They reached out and said, 'You know, we're really looking for some good use cases.' And I said, 'Well, we have some great ones. They're not necessarily sexy; I don't know if they're going to be on the front page of The Wall Street Journal. But they're going to make the lives of about 300 accountants 50 times better.'
From anomoly detection to margin analysis
The first Pilot Flying J Coleman app? Anomaly detection for finance. Clothier:
We let Coleman AI digest our accounting systems every morning, and our current monthly results, by store, by account, by line item - so you're talking hundreds of thousands of line items that no accountant could ever look through.
But poof, through the magic of whatever computer science and Coleman, 150-200 accountants stroll into work, and click a button. If they're responsible for the utility bills, or they're responsible for discounts, or whatever their line item is, it tells them, 'Hey, if I were you, I'd look at these eight items. And you may not have to look at anything else.'
As a result, they say, 'Okay, that saved me three hours; I'm probably not going to miss the thing that I was going to miss. I'm not going to be embarrassed when the CEO finds that thing I missed.'
Clothier wasn't joking about their (former) CEO Jimmy Haslam, who had a reputation for spotting a line item error amongst a stack of papers twelve inches thick. But as Clothier explained, the new CEO, Shameek Konar, is now an anomoly detection fan:
The other week, Konar asked me, 'Hey, come down here. Show me this anomaly thing.' And sure enough, I take him all the way through Infor Birst, and the P&Ls, and the anomaly detection. He's asking me, 'If we can do this, and - 'Everybody has this?' When I left the room, he was still playing with it.
And the fuel margin use case? To build it, Clothier says it took about 70 percent intellectual capital from his team. Infor was instrumental in developing the Coleman app, and determining the proper data sets to power it. What about those eye-watering projected numbers, increasing fuel margin accuracy to 99.99 percent, for an estimated revenue increase of $750,000 a year.
Is that realistic? Clothier says if you have the predictive precision, a penny here or there can add up, and add up massively. He says that given people's changing habits on how far they want to drive for gas (less in rush hour, perhaps longer on Saturday), Pilot Flying J should have precise price adjustments via Coleman AI - down to the daily and hourly level by store:
We might want to change prices up on Friday night. But we might want to make sure we're super-competitive come morning.
A location saves some pennies. And how do those add up?
In our business, and this is public knowledge, we're going to sell 7 billion gallons this year. Take a penny times 7 billion gallons. That's not $750,000 - it's $75 million... So that's why I say the number I threw out is exceptionally conservative. That's why pricing matters.
The wrap - AI Workforce Management on deck
Clothier's team is not stopping there: they're now expanding this capability to the entire P&L across the company. Clothier also sees an Infor Workforce Management use case:
As word has spread, the operations guys have a need in Workforce Management - Workforce Management will predict how many bodies you need to work. That's a big deal in retail. But it has to have sales projections to do it. It has to have some science behind it: 'What do you think we're going to sell tomorrow,' and now bring in the right number of people. Historically, we have not been very good at those projections. Now they're like, 'Well, could Coleman AI do that?' We believe it can. So we're in the middle of that project.
Clothier even wants to turn Coleman loose on Twinkies:
We're going to ship Coleman AI down to the Twinkie sale. By day, by time, lets it analyze and go, 'Okay, it's store one, here's what it looks like for the last, 18 months, whatever, by day, by hour, by transaction. Let's roll that up.'
Looks like Clothier is going to be busy for quite a while.
For more on Pilot Flying J's Coleman AI projects, check Infor's Innovation Showcase video.