The Cost of a Stockout: Why Availability Is a Product Problem, Not a Warehouse One

5 MINS

When a bestseller goes out of stock, most teams treat it as an operations failure, someone didn't reorder in time. After years of running replenishment, I've come to see it differently. A stockout is a forecasting decision that went wrong weeks earlier. By the time the shelf is empty, the mistake is already old. That's why I build EasyReplenish around one belief: availability is a product problem, not a warehouse one.

Stockouts and overstock are the same mistake

It's tempting to think of them as opposites, too little versus too much. They're not. Both come from the same root: guessing demand instead of forecasting it. Overstock is capital frozen on a shelf; a stockout is a sale handed to a competitor. One bleeds margin slowly, the other loses the customer instantly, and a brand chasing one usually overcorrects into the other.

The job isn't to avoid either extreme by being cautious. It's to make the demand signal sharp enough that you don't have to choose. That's the whole point of using 20+ data points to decide what to stock, when, and how much, every extra signal narrows the guess.

95% availability is a target, not a wish

At Udaan I was responsible for 60,000 products across six warehouses, with a hard line: 95% availability on bestsellers. A number like that changes how you think. You stop asking "did we reorder?" and start asking "what's our confidence on next month's demand for this SKU, at this location?"

Availability at scale is a portfolio problem. You can't protect every SKU equally, you protect the bestsellers ruthlessly and let the long tail flex. Knowing which is which, and updating that judgment as trends move, is most of the work.

Forecasting beats reordering

The instinct in supply chain is to react: stock dropped, so reorder. But reordering is always late by design, it starts the clock after demand has already moved. Forecasting starts the clock before. The difference between the two is the difference between chasing the market and meeting it.

That's why I'd rather invest in better sales-trend analytics than in faster reordering. A good forecast lets you reduce total inventory *and* hold availability at the same time, the two goals that everyone assumes are in tension actually move together once the prediction gets good enough.

What I'm building toward

EasyReplenish exists because most apparel and accessory brands are still making these calls on gut feel and last season's spreadsheet. They feel the pain, the dead stock, the missed bestsellers, but they don't have the signal to fix it. My job is to turn 20+ scattered data points into one clear decision: stock this, this much, now.

Get that right and the warehouse stops being where mistakes show up. It becomes where good forecasts quietly pay off.

Building Products for the Warehouse Floor, Not the Demo

Varun skipped presentations and built real AI products.

Varun Ravula was part of the April 2026 cohort at Curious PM, alongside 18 other talented participants.