Inventory

Stockouts start weeks before the listing goes inactive

A stockout is not the day the listing goes inactive. It is the four weeks before: velocity crept up, the reorder waited for a signature, the inbound shipment sat in a queue. By the time Amazon shows zero the sales are gone, the rank is slipping, and since this year the fee bill has already started climbing.

Days of supply is the number to watch

Amazon's low-inventory-level fee is charged per unit sold while a product's historical days of supply sits below the threshold on both the trailing 30-day and 90-day windows. In 2026 it reached more categories. It is triggered per variant, so one thin child ASIN generates fees on every sale even when the rest of the catalogue looks healthy.

That changes the economics of running lean. Thin inventory now costs you twice: in the sales you miss when you run out, and in the fee on every unit you sell on the way there.

Why averages fail

Reordering on last month's average velocity misses the season, the deal, the ad push and the competitor who just went out of stock. It also ignores that lead time is not one number: production, freight, and the inbound placement choice you make on the shipment all move it. Splitting a shipment to more destinations lowers the placement fee and lengthens the time until everything is receivable.

The reorder point has to move with velocity and lead time, weekly at least. A number set in January is wrong by March.

The reorder as a decision, not a spreadsheet

A good reorder is a proposal with its reasoning: the velocity trend and what is driving it, the lead time for this supplier right now, the placement option and what it costs, the cash it ties up, and the date the stock lands. The person deciding sees all of it on one screen and says yes, no, or less.

Then the shipment gets tracked to arrival, because a purchase order that leaves late is a stockout with a paper trail.

What AI changes here

Forecasting from daily patterns rather than monthly averages, across every SKU and marketplace, is work software now does well. It flags the stock risk weeks ahead, drafts the purchase order, and keeps the reorder point current as velocity moves. What it does not decide is how much of your cash to tie up. That is the operator's call, inside guardrails you set.

The result is quieter than it sounds: fewer emergencies, fewer air-freight bills, and a fee line that stops growing.

Back to the blogBack to home