Workshop Stock Takes: Why Parts Go Missing and How to Find Them

The stock take is not the painful bit. The painful bit is the twenty minutes afterwards.

Somebody counts bin A1-03 and finds seven oil filters. The system says eleven. Everyone in the room now has a theory. They went on jobs and nobody recorded it. They were never delivered in the first place. The last count was wrong. Somebody took a box to the other branch. One of them is probably right and there is no way to find out which, so the number gets adjusted to seven, everyone agrees to keep a better eye on it, and the same conversation happens next quarter.

The stock take found a discrepancy. It could not explain one. That is the actual problem, and it is not solved by counting more often.

Counting is easy, explaining is hard

Any system can hold a number and let you change it. What makes a count useful is being able to answer the follow-up question: not "how many are there" but "when did the other four leave, and on whose job".

That question has an answer only if every change to that number was recorded at the time it happened, with the reason attached. Most workshops do not have that, because stock is a single figure that gets edited. When somebody types 7 over 11, the 11 is gone. There is no record that it ever said 11, no record of who changed it, and certainly no record of why.

Edit a number often enough and it stops being evidence. It becomes somebody's opinion from an unknown date.

A ledger, not a number

The alternative is to treat stock the way accounts treat money: as a ledger of movements that are only ever added to, never edited.

In Vehiso, every change to a part's stock writes a row. The row records the quantity that moved, the cost, the balance that resulted, and the thing that caused it: a job card, a delivery, a transfer or a count. Rows are never rewritten. The current stock figure is not a number somebody typed, it is the result of everything that has happened to that part.

Which turns the awkward twenty minutes into something quite different:

A1-03 oil filter. Balance 11 on 2 September. Issued 2, job 4471, 4 September. Balance 9. Issued 1, job 4488, 9 September. Balance 8. Booked in 6, delivery from supplier, 11 September. Balance 14. Issued 4, job 4502, 16 September. Balance 10. Transferred 3 to the second branch, 18 September. Balance 7.

Count says seven. The system, read properly, also says seven. The eleven was somebody looking at the figure before the transfer.

Or the history genuinely stops short, in which case you have learned something specific and actionable: parts are leaving the shelf without going onto jobs, between these two dates, on this part. That is a process problem with a location, which is a completely different thing from "we seem to lose filters".

Who does the counting

The other reason stock takes get skipped is that they get treated as a management job, which means they happen when a manager has a free afternoon, which means quarterly at best.

They do not need to be. Counting is the easy half and the technicians are standing next to the racking all day. In Vehiso, counting a stock take is open to technicians. What stays with the parts desk or management is the half that needs judgement: reviewing the variances and deciding to apply them to stock.

That split matters more than it sounds. It means a count can be a Tuesday morning job rather than an event, and it means nobody adjusts a stock figure without somebody senior seeing what they are adjusting and by how much.

Counting smaller and more often

Once a count is cheap, the annual shutdown stock take stops being the only option.

Counting a bin, a shelf or a category, with the ones that move fastest counted most often, catches a drift while it is still four filters rather than a year of them. It also means the variance conversation is small enough to actually have. Nobody investigates a four-hundred-line variance report. Everybody will look into one part that is three out.

And because each count is itself a movement in the ledger, with its own date and its own author, the counts accumulate into a record of how accurate your stock has been over time. A part that is never out is a part you can stop counting. A part that is always out has something wrong with how it is handled.

What actually causes the gap

When workshops start explaining their variances rather than absorbing them, the causes turn out to be boringly consistent.

Parts that went on a job without being recorded. A technician grabs a clip, a bulb, a litre of oil. Small, individually trivial, adds up over a quarter.

Deliveries not booked in. The box arrived Thursday, went on the shelf, and the paperwork went in a tray. The shelf went up and the system did not.

Pack size substitutions. You ordered four and the factor sent a box of five, because that is how they come. Vehiso allows the over-receipt with a warning rather than refusing it, precisely because this happens constantly, but it has to be booked in as five for the ledger to stay true.

Transfers recorded at one end. Stock left one branch and never arrived at the other, on paper. Transfers in Vehiso are recorded at both ends as a matched pair for exactly this reason, and it is one of the themes in two branches, one shelf.

A previous adjustment. Somebody corrected a number once without a reason, and everything since has been measured from a fiction.

Every one of those is fixable. None of them is findable without a history.

If you hold almost no stock

Then do not do any of this.

A workshop that buys per job from a factor has nothing meaningful to count, and a stock take on a shelf of bulbs is time you will never get back. The part of a parts system worth switching on there is much smaller, and it is two fields on a job card line.

Stock takes are for workshops with racking. If that is you, the thing to take from this post is not that Vehiso does counts. It is that a count is only worth doing if a discrepancy can be explained, and explanation needs a history that nobody edited.