A stockout is the dashboard catching up, not warning you
The sequence is always the same. A SKU starts selling faster than forecast. Sell-through climbs. Somewhere around 80-90% depleted, someone notices and flags a reorder. By the time that flag gets raised, the demand that caused it has been building for weeks, and some real number of customers already hit a sold-out product page and left. Your dashboard reported the trend accurately. It just reported it too late to act on the customers you already lost.
This isn't a forecasting failure. It's a structural limit of the data. A sell-through number can only describe units that were actually available to sell. It has zero visibility into demand that showed up after the shelf emptied, and that gap is precisely the information a reorder decision needs most.
The signal that moves first
Before a stockout shows up as a sales spike, it shows up as talk. Someone asks where to find a size that's gone. Someone posts a photo asking if a color is coming back. Someone compares your product to a competitor's because yours wasn't available when they wanted it. None of this generates a row in your sales database, because none of it is a transaction. All of it is public, and all of it arrives earlier than the number it eventually explains.
The pattern that matters isn't one post. It's the same specific request, a size, a color, a variant, repeating across people who don't know each other, over a short window. One person asking is an anecdote. Ten people independently describing the same gap in two weeks is a reorder signal arriving while you still have time to act on it.
Why fashion feels this harder than most categories
Lead times are real and they don't move fast. Between a reorder decision and stock landing, weeks pass, sometimes a full production cycle. A brand that only reacts once sell-through data confirms a hit has already spent that entire lead time doing nothing, because the confirmation itself only arrived once it was almost too late to be useful. The brands that reorder calmly, ahead of the shortage, aren't forecasting better. They're listening to a signal that exists before the sales number does.
What to check before your next reorder cycle
- Pick one SKU currently trending toward sold out. Search for public mentions of it, or its size and color combination, from the last two to three weeks. Count how many are independent people versus the same handful of repeat customers.
- Check whether any of those mentions came after the product actually went unavailable. A cluster of "is this restocking" posts after a sellout is the clearest confirmation the demand kept going once the dashboard stopped counting it.
- Look at a SKU you decided not to reorder last season. See whether anyone was still asking for it months later. If they were, the sales data closed that door too early.
The dashboard tells you what already happened. The reorder decision needs to know what's about to.
Where this fits the bigger picture
This is the same asymmetry that shows up everywhere in Hugo, AI for fashion brands: your own sales data can only ever describe the products you already made and the stock you already had on shelf. It has no way to show you the demand that arrived after the shelf emptied, the demand for a variant you never carried, or the customer who compared you to a competitor and left. Hugo reads the public conversation continuously, so a reorder signal like this one surfaces while there's still lead time left to act on it, not after.