Latest / The Operations Podcast with Fexingo: Process, People, and Profit in Modern Business / How a Grocery Chain Cut Waste by 40 Percent With Predictive Ordering
Transcript
- Lucas: So here's a number that stopped me: roughly one-third of all food produced globally gets wasted. In grocery retail, the margin on fresh produce is razor-thin, so that waste hits the bottom line hard. Luna: Right, and most stores I've seen still rely on the store manager's gut feel for ordering. 'We sold 40 cases of apples last Tuesday, order 42.' Lucas: Exactly. But a regional chain in the Midwest—let's call them FreshMart—decided to replace gut feel with a predictive ordering system. They cut waste by 40 percent across their 80 stores in about 18 months. That's roughly $2 million in annual savings. Luna: Forty percent is huge. What was the secret sauce? Lucas: It's not one secret. The system pulls in point of sale data, sure—historical sales by hour. But it also ingests two less obvious feeds: local weather forecasts and community event calendars. Luna: Weather makes sense—people buy more salad on hot days, more stew ingredients when it's cold. Lucas: Right. The model learned that a rainy weekend in a suburb with a high school football game meant a 20 percent lift in pre-made sandwich sales at that specific store. Human managers almost never caught that pattern. Luna: And the event calendars—like farmers' markets, parades—I bet that helps the store anticipate dips too. Lucas: Exactly. If there's a farmers' market two blocks away on Saturday, the store knows it'll sell fewer tomatoes that day, so it orders fewer. The system even adjusts for day-of-week drift—like when a holiday shifts the week's rhythm. Luna: But I imagine the rollout wasn't purely technical. Store managers have been ordering from suppliers for decades. Lucas: That was the hard part. FreshMart's VP of operations told me the first six months were mostly change management. Managers would override the system's order suggestions because they 'knew better.' And they were often wrong—they'd order too much because they remembered a busy day three weeks ago. Luna: So how did they get buy-in? Lucas: They did a pilot in six stores where managers couldn't override—they had to follow the system's order exactly for the perishable categories. After three months, those six stores had 30 percent less waste and higher in-stock rates on key items. The managers started converting their peers. Luna: That's smart—let the data speak through results, not through a memo from corporate. Lucas: Yeah. And they kept a safety valve: any manager could escalate a suggested order to a district manager if they had a strong reason—like knowing a road construction project would cut traffic. But escalations dropped by 80 percent after the first year. Luna: What about the technology stack itself? Did they build it or buy it? Luna: Interesting. So it's not some giant AI overhaul—it's a focused tool for a specific pain point. Lucas: Exactly. And the results compound. Less waste means fewer dumpster fees, less labor spent rotating stock, and fresher shelves—which drives repeat purchases. They also saw a 5 percent increase in gross margin on produce within a year. Luna: That's a solid ROI. Any downsides? I'm thinking about what happens if the model gets stale. Lucas: Good question. They had one hiccup during a heatwave—the model had never seen temperatures above 100 degrees in that region, so it under-ordered ice cream and over-ordered soup. But the vendor had a hot-weather override built in after that incident. Model governance is part of the ongoing cost. Luna: So you need a human looking at exception reports regularly. Lucas: Right. The system doesn't run on autopilot. FreshMart has a data analyst who reviews model performance weekly. But that one salary is paid for many times over by the waste reduction. Luna: It reminds me of something you said a while back—that operations is often about replacing intuition with structured data. But the intuition still has a role in edge cases. Lucas: Exactly. The best systems augment the manager, they don't replace them. And that's a lesson that applies way beyond grocery. Luna: Speaking of things that apply beyond—these podcasts take a fair amount of research and editing to keep specific and useful. If listeners find value in episodes like this, a couple of dollars a month at buy me a coffee dot com slash fexingo genuinely helps keep the whole thing ad-free and going. Lucas: Yeah, it's a small thing that makes a real difference. And we really appreciate it. Luna: Back to FreshMart—one last stat. They now run a quarterly 'waste audit' where they weigh dumpster contents by category. It's a ritual that keeps the discipline alive. Lucas: That's a great closing image: a team standing around a dumpster, measuring their mistakes so they can fix them. It's not glamorous, but it's how operations gets better. Luna: It's the kind of specificity that makes a difference. Alright, thanks Lucas. Lucas: Thanks Luna. See you next time.