Your System Says Twelve. The Shelf Says Nine.

The meeting that goes nowhere

Somebody finds the gap. Twelve on the system, nine on the shelf. There's a short meeting. The conclusion is that people need to be more careful, and everybody agrees, and nothing changes, and three months later it's a different part and the same meeting.

The meeting goes nowhere because the diagnosis is wrong. Almost none of your variance is carelessness, and less of it is theft than you think. Most of it is a timing problem: something happened on your floor at eleven in the morning, and the system found out about it at five in the afternoon, or on Thursday, or never.

Where the three units went

In a parts business, the same handful of gaps account for most of it.

The part that left before the invoice. A fitter needs it now. The job is on the hoist, the customer is waiting, and the part goes to the bay on the understanding that it'll be written up afterwards. Most of the time it is. The variance is built out of the times it wasn't.

The delivery counted at the door. Twenty boxes arrive, twenty boxes are signed for, and one of them has nine units in it instead of ten. You didn't receive twenty boxes. You received a quantity, and nobody checked it against the line.

The return that went back on the shelf. A customer brings a part back. It physically returns to the bin, because that's the obvious place for it. Whether it returned to the stock record depends on who was on the counter.

The same part under two codes. The supplier's number, the OEM number, and the number your business used before you changed suppliers in 2019. Three records, one physical part, and stock spread across all three in a pattern nobody can reconstruct. This one is quietly enormous in auto parts, and it doesn't look like variance — it looks like you're overstocked on one code and out on another.

The write-off nobody wrote. Something got damaged. It went in the bin at the back. It is still on the system, and it will be on the system at the next count, and then it will be a variance.

Notice what these have in common. In every one of them, a real event happened and a person was expected to remember to record it separately, later, using a different tool than the one in their hands. That's not a discipline problem. That's a design problem.

Why "be more careful" fails

Because you're asking people to do two jobs where the business only pays for one.

The fitter's job is to get the car off the hoist. The counter salesman's job is to serve the person in front of them. Recording the stock movement is a second task, done afterwards, with no immediate consequence for skipping it and a very immediate consequence for slowing down. When the shop is busy, the second task loses. It will always lose, in every business, no matter how good the people are.

The only durable fix is to stop having two tasks. If the act of taking the part is the act of recording it, there's nothing left to forget.

What a scan actually does

It collapses the event and the record into one thing that happens at the same moment, in the same place, by the same person.

That's it. That's the whole mechanism. A scanner isn't clever — it's a device for making the record happen at the point the stock physically moves, instead of at a desk later. Everything else follows from that.

What follows is worth having:

• The movement has a timestamp, so "when did we last have twelve?" has an answer.
• It has a person, so a pattern in the variance has somewhere to point.
• It has a direction — received, picked, returned, written off — so you can see whether you're leaking on the way in or on the way out.

Most businesses discover their leak is on the way in. That surprises people. Receiving is boring and nobody watches it, and it's where the discrepancy gets built into the record before the part has been on your shelf for five minutes.

Knowing you have it isn't the same as finding it

There's a second half to this, and it's the one that actually costs you sales.

Suppose the count is right. Twelve on the system, twelve on the floor. Your counter salesman still has to find them, and "somewhere in the back" is not an address. Two minutes of looking with a customer waiting is a long two minutes. Ten minutes, and the customer is phoning the competitor from your parking lot.

Bin locations fix this, and they're cheap. Label the location as well as the part, tie the two together with a scan, and the question changes from "do we have it?" to "it's in C2-04." A new employee can find stock on their first morning. That's worth more to most parts businesses than the accuracy gain, and it's the thing they didn't expect to get.

The honest limits

A scan records what was scanned. If someone takes a part without scanning it, the system is as blind as your spreadsheet was. What changes is the incentive: scanning becomes the fastest way to do the job, so it's the path people take when nobody's watching.

And none of this repairs history. Your first count after labelling will throw up variances that have been accumulating for years, and some of them will be ugly. That's not the system failing. That's the first time you've been able to see the number — and it gets steadily less interesting every count after.

The test worth running

Don't take a vendor's word for any of this, ours included.

Pick one aisle. Label it, scan it, and run it properly for a month while the rest of the business carries on exactly as it does today. Then count that aisle and count a comparable one you didn't touch. The difference between those two numbers is the only argument that matters, and it's yours, from your floor.

That's what we bring the kit for.