Moving inventory checks off paper: replace the transcription, not the walk

01 / FIELD NOTE

Keep the decision tied to the operating context.

A paper-based count gets criticised for being slow, and its speed is rarely the deciding problem. The people doing it are moving through the stock, looking at it, and writing down what they see, which is the same work any other method has to do. What makes paper expensive is everything that happens to the numbers after they are written.

The first cost is the handover. A written count becomes a typed one, and the typing is a second opportunity to be wrong with no way to tell the two errors apart afterwards. A digit transposed, a column misread, a line skipped because the handwriting ran two entries together — each produces a discrepancy that will be attributed to the count rather than to the transcription.

The second is latency, and it is structural rather than accidental. Paper can only be entered when someone has the sheet and the time, so the record is updated after the count rather than during it. Everything that happens in between is transacted against a figure that is already known to be stale, and the longer the sheet sits, the more decisions are made against it.

The third is that a sheet records a number and nothing else. It does not reliably record when the count was taken, who took it, which location it covered, or what condition the goods were in. Those are exactly the details needed to investigate a discrepancy later, and they are the details that paper loses first because there is nowhere on the form to put them.

The fourth is that paper cannot represent an exception. A form that asks for a quantity forces every observation into a number, so a shelf that was empty, a location that was inaccessible and an item found in the wrong place all get written the same way — or not written at all. The information that would explain the discrepancy is discarded at the moment it is observed.

The fifth is that paper is invisible until it is collected. Nobody can see how a count is progressing, which areas have been covered and which have not, or whether the count is going to finish. Problems are discovered at the end, when correcting them means repeating work rather than adjusting it.

So the first change worth making is not the counting method but the capture. A handheld that records the count as it is taken removes the transcription handover, timestamps and attributes each entry, holds the location it belongs to, and can accept an exception instead of forcing a number. That is a smaller change than it sounds and it addresses the largest single source of error.

Capture as it happens also makes the record live rather than eventual. If the device updates the system as it goes, a count becomes a process that the count itself can act on: an area that is showing an unexpected variance can be re-checked while the person is still standing in it, which is when a re-check costs almost nothing.

Bulk reading is a further step, and it changes the rhythm rather than the accuracy. Reading many tags in a pass makes a count fast enough to be routine, which is what allows counting to become frequent and partial instead of rare and total. The gain comes from the frequency, not from the speed of any individual count.

The order matters. Moving from paper to a handheld improves the quality of a count that is still occasional. Moving from occasional to frequent counting is what reduces drift, and it only becomes affordable once the count is fast. Attempting the second change without the first produces frequent counts of the same unreliable data.

Not everything can leave paper at once, and the exceptions are worth planning for. Locations where the radio performs badly, goods that are not tagged, items inside sealed packaging, and the awkward cases that a person has to look at individually. A process that has no way to record those will either skip them or record them wrongly, and both are worse than a defined manual fallback.

The measure of the change is whether the record can be trusted without a second opinion. If supervisors still keep a private sheet for the categories they care about, the paper was not the problem and removing the form did not address it. The useful test is whether a discrepancy can be traced to a cause from the record alone — and that is what the capture change, rather than the paper change, is for.