Latest / The Operator's Manual with Fexingo: Daily Decisions, Tactical Wins, and Business Execution / How One Grocery Chain Cut Produce Waste 39 Percent With One Forecast Change
Transcript
- Lucas: Back in early 2023 a regional grocery chain in the Midwest was throwing away roughly 18 percent of its fresh produce every single month. That means almost one out of every five avocados, heads of lettuce, and cartons of strawberries never made it to a shopping cart. Within six months they cut that number to 11 percent — a 39 percent reduction — by changing exactly one thing: how they forecasted demand at the individual store level. Luna: So not new equipment, not new coolers, not a new supplier relationship — just a smarter way of using data they already had. Lucas: Exactly. And that's the spirit of this whole series — one targeted operational tweak that creates outsized impact. The chain had been using a classic weekly replenishment model: each Monday, the central warehouse would push out a standard allocation based on the previous week's sales, adjusted for the time of year. Luna: Right, the 'set it and forget it' approach — which works fine for canned goods but falls apart for things that go bad in days. Lucas: Precisely. The problem is that produce demand is incredibly local and incredibly volatile. A store near a college campus might sell three times more bagged salad during exam week because students aren't cooking. A store next to a park sees a massive spike in watermelon sales on the first 80-degree Saturday of spring. The weekly forecast simply couldn't catch those patterns. Luna: So what did they swap in instead? Lucas: They built a simple daily forecast model — and I mean simple. It started as a spreadsheet. For each store they took three years of daily sales data for every produce category, then layered in two external signals: the local weather forecast — specifically high temperature and precipitation probability — and a calendar of local events, like high school football games, farmers markets, and municipal holidays. Luna: And they did this for every store? How many stores are we talking about? Lucas: Roughly 80 stores. They didn't hire data scientists. The model was essentially a weighted average: it looked at what that store sold on similar days in previous years — same day of week, similar weather, similar event profile — and then applied a confidence interval. Store managers could override if they had local knowledge, but overrides became rare after the first month. Luna: So the spreadsheet essentially said: 'Store 47, Tuesday looks sunny and 75 degrees, there's a Little League tournament in the neighborhood — bump up your watermelon order by 40 percent.' Lucas: Exactly that. And the results were dramatic. Within 90 days, the chain's overall produce shrink dropped from 18 percent to 13 percent. By month six it was 11 percent. That's a 39 percent reduction in waste. But here's the kicker: sales actually went up too, because they were less often out of stock on high-demand days. Luna: So they sold more and wasted less — that's the kind of double win that makes CFOs very happy. Lucas: Right. The company estimated that the improvement in gross margin on produce alone added roughly 1.2 million dollars to annual profit across the chain. And the total cost to implement? A few hundred hours of a supply chain analyst's time, plus some spreadsheet training for store managers. Luna: That's an incredible return on a very small investment. What about adoption — were store managers resistant to trusting a spreadsheet over their gut? Lucas: Initially, yes. A few managers said things like 'I've been doing this for 20 years, I know what my customers want.' But the chain did a smart thing: they ran a two-month pilot in 10 stores, and they let the managers see the spreadsheet's predictions side by side with their own orders, without forcing them to follow it. By the end of the pilot, those managers were voluntarily adopting the forecasts because they saw the waste numbers drop in real time. Luna: That's the classic 'show, don't tell' approach to change management. So the spreadsheet became the default, but managers could still override — and over time they learned to trust it. Lucas: Exactly. And eventually, the chain rolled it out chain-wide. They also found that the model worked best for high-turnover items like berries, bagged salads, and fresh herbs — things with a shelf life of three to five days. For longer-life items like apples or potatoes, the improvement was smaller, but still measurable. Luna: That makes sense. Faster-moving, shorter-life items are where the volatility hurts most. So if a listener is running a similar operation — a restaurant, a bakery, even a farm stand — what's the one takeaway here? Lucas: The takeaway is that you don't need a machine learning platform or an expensive ERP upgrade to make a big dent in waste. You need three things: daily granularity, external signals that actually affect demand, and a system that makes it easy for the person placing the order to see what the data suggests. In this case, it was a spreadsheet. That's it. Luna: And the chain is still using that spreadsheet today? Or did they eventually graduate to something fancier? Lucas: They did eventually migrate it into their warehouse management system — but that was mostly for convenience, not necessity. The core logic remained the same. The spreadsheet was doing the job. It just wasn't scalable once they started thinking about adding more stores or more categories. Luna: It's a nice reminder that operational excellence often comes from a simple insight applied consistently, not from buying the shiniest new tool. Lucas: Absolutely. And if you're in a position where you're trying to justify a similar change to your boss or your board, you can point to that 1.2 million dollar number — and the fact that it came from a spreadsheet, not a capital expenditure. Luna: Now, something else I want to touch on — you know, we do this show because we genuinely enjoy digging into these stories and sharing them. And a couple of dollars a month from listeners who find value in it is what keeps the podcast going without ads. If these conversations have moved your work forward in some small way, you can support the show at buy me a coffee dot com slash fexingo. Lucas: Yeah, we've had listeners tell us they saved tens of thousands of dollars by applying something they heard on an episode like this one. That's why we do it. And every bit of support helps keep it ad-free and focused on the tactics. Luna: Exactly. So back to the grocery chain — I'm curious, did they ever apply the same daily forecast logic to other perishable departments, like meat or dairy? Lucas: Great question. They started with produce because the waste was most visible — literally rotting in the back room. But once the produce model proved itself, they rolled out similar daily forecasts for the deli and bakery counters, which also have short shelf lives. The deli saw about a 25 percent reduction in waste within three months. The bakery was more mixed because a lot of their baked goods have a same-day sale window, but they still improved. Luna: So the pattern holds: find your most perishable, most volatile category, fix that first, and then propagate. Lucas: Exactly. And the other lesson is humility: the chain's leadership admitted that the old system was fundamentally flawed because it treated every store as if it faced the same demand patterns. They knew stores were different, but they didn't have a mechanism to account for it. The daily forecast forced that localization. Luna: It's almost like they discovered that their biggest operational problem wasn't in the supply chain — it was in the assumption that one size fits all. Lucas: Right. And once they broke that assumption, the solution was almost obvious. But it took a pilot, some data, and a willingness to let a spreadsheet challenge years of intuition. Luna: So if you're listening and you're thinking, 'I wonder how much waste we could cut by just looking at our data differently,' that's the challenge — go find your version of the daily forecast. Lucas: That's the operator's mindset. Look at your biggest operational leak, and ask: 'What would happen if we updated our plan every day instead of every week?' The answer might surprise you.