# Gotchas The mistakes that stock out a bestseller or bury cash in a pallet. Inventory errors are slow and expensive — they show up as a missed sale or a shelf nobody touches for two years. --- ## Gotcha: counting a stockout as zero demand **Why it matters:** This is the most damaging error in the whole skill, and it is self-reinforcing. The item ran out because it sells. The zero-sales days drag the velocity down. The next order is smaller. It runs out sooner. Repeat. ### Bad ``` Merv 11 filter: 12 units sold in 28 days = 0.43/day. Slow mover, no reorder. ``` (the shelf was empty for 20 of those 28 days) ### Good ``` Merv 11 filter: 12 units in the 8 days it was actually in stock = 1.5/day. Out of stock 20 of the last 28 days — that is suppressed demand, not slow sales. Planning on 1.5/day. ``` --- ## Gotcha: ordering into a one-week spike **Why it matters:** A single bulk order makes an ordinary item look like a runaway for exactly seven days. Sixty days of cover sized off that number is a pallet the owner paid for and will look at for a year. ### Bad 7-day velocity 5.8/day vs 28-day 0.4/day. Size 60 days at 5.8/day. ### Good ``` 7-day rate is 14x the 28-day rate. Looking at the orders: one sale of 40 units on the 9th. That is a job, not a trend. Sizing on the 28-day rate and noting it in case that customer comes back. ``` --- ## Gotcha: recommending a reorder for a zero-velocity item **Why it matters:** Zero sales in 28 days means stop buying. A reorder point set years ago in the ERP will happily say otherwise, and following it converts working capital into shelf decoration. ### Bad NetSuite reorder point says 24, on hand is 6, so order 18. ### Good ``` Zero units sold in 28 days. The stored reorder point says 24 — that is stale. Not a reorder. Six units, USD 340 tied up. Worth clearing. ``` --- ## Gotcha: forgetting what is already on order **Why it matters:** Two orders for the same item three weeks apart is the classic double-buy, and it usually happens right after a stockout scare when the owner is paying attention. ### Bad On hand 38, target 258, order 240. ### Good ``` On hand 38, 160 already inbound from the 3/2 order, target 258 total. That leaves 60 — 2 cases of 40, so 80 units. If that inbound number is wrong tell me, it changes this. ``` --- ## Gotcha: a stock count with no as-of date **Why it matters:** Everything downstream is stock divided by velocity. A two-week-old count on a fast mover produces a stockout date that has already passed. ### Bad Take the uploaded count at face value and project from today. ### Good Ask when it was counted. If the count is from the 1st and today is the 15th, either net out two weeks of sales or ask for a fresh count on the fast movers, and say which you did. --- ## Gotcha: seasonality invented from partial history **Why it matters:** A multiplier computed from nine months looks exactly like one computed from three years. It gets used to buy real stock, and nobody can tell later which kind it was. ### Bad ``` June factor: 1.6x (computed) ``` (from 120 days of data that do not include a June) ### Good ``` 120 days of history — no June in there, so I cannot compute a season. You know this better than the data does. Roughly how much heavier is June? ``` --- ## Gotcha: velocity at the product level instead of the variant **Why it matters:** A product that sells well in one size and never in another looks healthy in aggregate. The good size stocks out while the bad size accumulates, and the product-level number shows neither. ### Bad Filter, 20x25: 4.1/day, 38 on hand, healthy. ### Good Split by variant. The 1-inch is running out in four days; the 4-inch has nine months of cover. Two different problems, one product line. --- ## Gotcha: reporting units without dollars **Why it matters:** Owners decide about cash. "Order 240 filters" is not a decision they can make; "USD 888" is. ### Bad A buy list of quantities with no cost column. ### Good Every line costed, every vendor subtotaled, and the total stated before the approval question. --- ## Gotcha: quietly accepting a vendor substitution **Why it matters:** A different part number goes onto a customer's equipment. That is the owner's call and sometimes a warranty question, never a shipping detail. ### Bad Vendor swaps a compatible capacitor, order updated, nothing said. ### Good Report the substitution, the part numbers both ways, and the effect on cover if the owner rejects it. Then let them decide.