Quick answer: A zero-units alert fires only after the damage is already happening. Days of supply (units divided by daily velocity) is the number that actually warns you before a stockout costs weeks of lost rank and Buy Box share.
Most inventory alerts are built around one moment: the count hitting zero. That's the easiest thing to detect, so it's what most tools fire on. It's also close to the least useful moment to find out, because a stockout doesn't start costing you anything the instant the count hits zero. It starts costing you days before, and it keeps costing you for a while after you've restocked.
Days of supply is the number that actually predicts trouble
Days of supply, current units divided by average daily sales velocity, is the metric that tells you something's coming before it arrives. A listing sitting at 40 units with 2 units per day of velocity has three weeks of runway. The same 40 units against 8 units per day of velocity has five days, and if lead times run anywhere close to the 8 to 10 week China-to-FBA average a lot of sellers are dealing with right now, five days of runway against an eight-week replenishment cycle isn't a scheduling inconvenience. It's a stockout that's already locked in, whether or not anyone notices yet.
Zero-units alerts can't see any of that coming. They fire at the exact moment it's too late to do anything but wait, which makes them closer to a postmortem notification than an actual warning.
The cost isn't just the missed sales during the gap
A stockout of even three to five days can produce a BSR drop that takes weeks, not days, to recover from, because Amazon's ranking algorithm reads the sales interruption as reduced demand rather than as a supply problem. It takes sustained sales afterward to rebuild the rank the listing had before the gap opened up. The same pattern applies to Buy Box allocation, which doesn't fully snap back the moment inventory reappears. There's a recovery lag layered on top of the outage itself, and that lag is rarely accounted for when sellers estimate what a stockout actually cost them.
So the real cost of a stockout isn't just however many units didn't sell while inventory sat at zero. It's that gap, plus a slower-selling period afterward while rank and Buy Box allocation claw back to where they were, often adding up to more lost revenue than the visible stockout window itself. A five-day stockout followed by two weeks of depressed sales while rank recovers is a very different number than five days of lost sales alone, and it's the number most sellers never actually calculate.
A newer wrinkle: Amazon now penalizes running too lean, too
The incentives here aren't purely "hold more stock, avoid stockouts" anymore. Amazon's low-inventory-level fee, tracked alongside your Inventory Performance Index, kicks in when days of supply drops below a set threshold, and that threshold moved from 28 to 35 days in 2026, meaning the safety margin sellers used to plan around got smaller without most sellers being aware the rules had shifted underneath them. Sit too lean now and there's a fee stacked on top of the stockout risk. Sit too heavy and long-term storage surcharges start applying at 181 days. The comfortable middle got narrower on both sides in the same year, which makes the whole balancing act less forgiving than it used to be.
This matters more for sellers managing large, varied catalogs than it does for a seller with a handful of SKUs. With ten products, checking days of supply by hand every Monday is realistic. With a few thousand active titles, the odds that any single one slips from comfortable to critical between checks go up substantially, simply because there's more surface area for something to be missed.
What this actually requires watching
None of this is answered by a binary in-stock or out-of-stock signal. It requires an actual number, days of supply, computed from real velocity, per ASIN, checked often enough that a listing sliding from three weeks of runway down to five days doesn't go unnoticed until it's already sitting at zero. It also requires that number to update as velocity changes, since a title that suddenly starts selling faster can burn through what looked like a comfortable buffer far quicker than a static "reorder point" set weeks earlier would suggest.
This is what EcomSanity's Monitor table surfaces directly: days of stock computed from real sales velocity per ASIN, so a slow leak shows up on the dashboard days before the listing actually goes dark. For the full formula, sales-window tradeoffs, and how to handle one-off inventory, see days of inventory for Amazon sellers. For what to actually do once the countdown runs out, see the Amazon stockout recovery plan.
Frequently asked questions
Why is a 'zero units left' alert too late?
A stockout doesn't start costing money the moment the count hits zero, it costs money in the days before, and Amazon's ranking algorithm reads the resulting sales interruption as reduced demand, so recovery lags for weeks afterward.
What's the low-inventory-level fee?
A fee Amazon charges when days of supply drops below a set threshold, which moved from 28 to 35 days in 2026, narrowing the safety margin sellers plan around.
What metric should I track instead of a binary in-stock or out-of-stock flag?
Days of supply, computed from real recent sales velocity per ASIN, updated as velocity changes rather than a static reorder point set weeks earlier.