Why you're 98% accurate and still can't find anything

It's the most common sentence in the complaints we've collected, and it's usually said with a shrug: our inventory accuracy is fine, we just can't find things. Both halves are true at the same time, and that's not a contradiction — it's a measurement problem.

You're measuring quantity accuracy: does the system say 40 when there are 40? That number is probably genuinely good. What you're not measuring is location accuracy: does the system know which shelf those 40 are on? Almost nobody tracks that separately, so almost nobody knows how bad it is.

Measure it, this week

Pick twenty random SKUs. For each one, ask the system where it is, then walk over and look. Score it three ways:

• Right quantity, right location — pass.
• Right quantity, wrong location — fail. This is the one you're currently not counting.
• Wrong quantity — fail.

Operations that believe they're at 98% typically land somewhere in the 70s or 80s on this. That gap is your problem, quantified. It takes about forty minutes and costs nothing.

Where the gap comes from

In the complaints we tagged, it's almost always the same story, and it isn't carelessness. The system says put it in A-12-C. A-12-C is full. The person putting it away is holding a box, the truck is still being unloaded, and there's a decision to make in about four seconds. So it goes in A-14, which is empty, and the intention is to tell someone later.

Nobody writes it down, because there's nowhere to write it down that anyone reads.

That single unrecorded decision is now a permanent lie in your records. Your quantity is still perfect. Your location is wrong forever, or until someone stumbles on it.

The fix has three parts, and only one involves software

First: give put-away an explicit override path. On paper this is one column on the sheet — actual location — filled in every time, even when it matches. In software it's a field that doesn't block you.

Second: whoever closes the day reconciles those overrides. Ten minutes. This is the part people skip, and it's the part that makes the rest work.

Third, and this is the one that decides whether any of it survives: never punish anyone for using the override. The moment reporting a deviation gets someone told off, people stop reporting deviations. They don't stop deviating — the truck still arrives, the bin is still full. You just go blind.

Why this comes before choosing software

We think this comes before choosing software, and we'll be straight that it's a belief and not something our data proves: locations labeled and standardized before any system goes in matter more than which system it is.

Software that receives good location data makes it better. Software that receives guesses makes them faster and more confident. If your put-away is honest, almost any tool will work for you. If it isn't, no tool will.