Latest / The Operator's Manual with Fexingo: Daily Decisions, Tactical Wins, and Business Execution / How a Single Pricing Error Cost 10 Million in One Week
Transcript
- Lucas: So there's a moment in every operator's life where you look at a number and think — that can't be right. And sometimes it isn't. But sometimes the number is wrong for a full week before anyone blinks. Luna: What's the damage when it takes that long? Lucas: Well, in a case I was tracking from a mid-size retailer — let's call them Coastal Home Goods, about $400 million in annual revenue — a single pricing error on two hundred SKUs cost them roughly ten million dollars in lost margin over seven days. Luna: Ten million from one upload? That's eye-watering. Lucas: Right. And it wasn't a sophisticated hack or a malicious actor. It was a CSV file. Someone in merchandising exported a pricing table, made a bulk adjustment in Excel, and when they re-uploaded, a column misalignment caused all two hundred high-margin items — think premium bedding, specialty cookware — to be priced at cost instead of the usual 2.5x markup. Luna: So the items were selling, but at zero margin. Lucas: Exactly zero. The website showed the correct retail price in the product description, but the cart price—the one that actually charges the customer—had somehow defaulted to the cost column. So every unit sold generated revenue equal to what the company paid for it. No contribution to overhead, no profit. Luna: And it took a week to notice. What finally caught it? Lucas: A night-shift inventory analyst named Maria — she was reconciling end-of-day gross margin percentages and saw the home goods category had dropped from 58 percent margin to 41 percent in a single day. She flagged it, escalated to pricing, and within two hours the error was corrected. But by then, the company had sold about forty thousand units at zero margin. Luna: Forty thousand units times the margin they lost — that's about two hundred fifty dollars per item on average. Quick math says ten million. Lucas: That's the number. And here's the part that keeps operators up at night: the error was introduced on a Tuesday afternoon. The first automated margin report ran Wednesday morning, but nobody looked at it until Thursday because the category manager was on vacation. By Friday, the volume had spiked — the items were effectively priced 60 percent below normal retail, and the algorithm started promoting them. Luna: Oh, so the system saw high conversion and low price and started featuring those products on the homepage. Lucas: Exactly. The pricing engine treated the discounted price as intentional. So by Saturday, the two hundred mispriced SKUs accounted for 18 percent of total site revenue — all at zero margin. The company essentially paid to give away its best products. Luna: What's the structural lesson here? Because anyone with an e-commerce operation could face this. Lucas: The structural problem is that the company had three layers of automation but no layer of human pattern recognition until the margin report. The CSV upload had a validation step that checked for format — correct number of columns, proper data types — but it didn't check for semantic reasonableness. It didn't ask: 'Hey, is it normal for two hundred items to suddenly have a 60 percent price drop?' Luna: So the system was checking syntax, not sense. Lucas: Right. And after this incident, Coastal Home Goods added what they call a 'margin guardrail' — a simple rule that flags any upload where more than ten SKUs change price by more than 20 percent in one direction. That rule alone would have caught the error within minutes of the upload. Luna: That seems cheap to implement. A few lines of code. Lucas: Cost them maybe a day of engineering time. And it now prevents what could be a recurring ten million dollar leak. But here's the thing — they had a similar rule in their physical retail system. If a store manager changed the price of more than five items by more than 15 percent, the system required a supervisor override. They just never ported that logic to the e-commerce platform. Luna: So the operational wisdom existed in one channel but not the other. That's a classic organizational blind spot. Lucas: Classic. And it's worth noting that the cost of this error wasn't just the lost margin. There was also the customer impact. Some customers bought the items at the correct price earlier in the week, then saw the same product 60 percent cheaper later. The company had to issue refunds for the difference to avoid PR backlash — that added another half a million in goodwill adjustments. Luna: So the total bill was closer to ten and a half million. Lucas: And some reputational noise. But the operational fix was straightforward: cross-channel parity on pricing controls, and a human in the loop for any bulk change above a threshold. Luna: It reminds me of something I read about Toyota's production system — they have an 'andon cord' that any worker can pull to stop the line if they see a defect. Maria pulling that margin report was essentially pulling the andon cord. Lucas: That's a perfect analogy. And the key is that Maria had the authority to escalate without going through three layers of management. She didn't need permission to be curious. That's a cultural element — if she had been told 'just log the variance and wait for the category manager,' the error might have run another week. Luna: So the fix had two parts: the technical guardrail, and the cultural permission to flag anomalies. Lucas: Exactly. And the technical guardrail alone would have been insufficient if the margin report had been ignored. But the combination — a system that catches dramatic price swings, plus a team that's empowered to act on outliers — that's what stops a ten million dollar bleed. Luna: Let's talk about the broader pattern. How common are these pricing data errors? Lucas: More common than most retailers admit. A 2024 study by a pricing analytics firm found that roughly 1 in 400 price changes in large e-commerce catalogs contain an error of more than 30 percent. That doesn't sound like much, but for a catalog of 100,000 SKUs, that's 250 potentially catastrophic changes per cycle. Luna: And most companies don't catch them until the next reconciliation — which might be weekly or monthly. Lucas: Right. And when you're dealing with thin margins — say 5 to 10 percent net — a single pricing error can wipe out a week's worth of profit from an entire category. Coastal Home Goods was lucky it was only a week. I've seen cases where errors ran for a full month before anyone noticed. Luna: What about pricing psychology? Some companies intentionally use loss leaders. How do you distinguish an intentional promotion from an error? Lucas: Great question. The key is metadata. Intentional promotions are typically logged with a campaign ID, a start and end date, and an expected margin impact. Errors have none of that. So the guardrail should check not just the magnitude of the price change, but whether it's associated with a known campaign. If there's no campaign ID, flag it. Luna: That's a simple rule but it requires that the promotion process be disciplined — that every discount is entered through the proper channel. Lucas: Exactly. And that's where a lot of companies trip up. Someone on the merchandising team has a spreadsheet with 'temporary price adjustments' that they upload directly without going through the promotions module. That's how errors slip in. Luna: So the operational fix is both technical and procedural: enforce the promotions workflow, and add a margin guardrail for anything that bypasses it. Lucas: That's the playbook. And it's worth noting that the cost of implementing these controls is trivial compared to the potential loss. A one-time engineering investment of maybe ten thousand dollars, versus a recurring risk of ten million per event. Luna: If today's episode was useful to you, this kind of tactical deep dive is exactly what listener support keeps ad-free. You can help at buy me a coffee dot com slash fexingo. Lucas: Yeah, that really does make a difference. So back to the margin guardrail — what else can operators learn from this? Luna: I think the biggest takeaway is that the most expensive errors often come from the simplest mistakes. A column shift in a CSV file is as basic as it gets. But without a system that checks for reasonableness, it can go undetected for days. Lucas: And the human layer — Maria checking the margin report at the end of her shift — that's the final safety net. Technology can catch the syntax, but only a person who knows what 'normal' looks like can catch the anomaly. Luna: So what's the one thing you'd recommend a mid-size retailer do tomorrow? Lucas: Implement a margin guardrail on any bulk price upload. Threshold: if more than ten SKUs change by more than 20 percent, require a second approval. That's it. That one rule would have saved Coastal Home Goods ten million dollars. Luna: And empower your inventory analysts to flag anomalies without waiting for permission. Lucas: Exactly. Maria's action was the real hero here. Systems fail. People catch the failures, but only if you give them the tools and the trust.