There's a count variance: is it an error or is it waste?

The count finishes, the report lands, and one line comes up short. The first question in the meeting is always the same: “where did this variance come from?” The answer is usually a guess — maybe it was missed during counting, maybe it was mistyped at entry, maybe something broke. Because it is a guess, nobody can act on it.

There is one structural reason a variance cannot be explained: some of the legitimate ways stock leaves are never recorded at all. The two most commonly skipped are waste and consumption.

Waste and consumption are not the same thing

Both reduce stock, but they happen for different reasons, and when they are conflated both become invisible.

• Waste: broken, spoiled, expired, or damaged goods. An unplanned loss, and the goal is to reduce it.
• Consumption: material used in production, assembly, or maintenance. A planned draw, and which job it went to should be known.

When neither is recorded, stock quietly erodes. The number that appears at count time is really the sum of unrecorded outflows accumulated over months — but the count shows it as a single line on a single day. That is why the cause stays invisible.

The real cost of an unexplained variance

The real loss here is not the monetary value of the missing goods. The real loss is that you cannot fix anything, because you do not know the cause:

• If the waste rate is high, nobody knows which product, which rack, or which shift it is high on.
• Without consumption records, the true material cost of a job cannot be calculated.
• The variance recurs systematically but is closed each time as “a counting error”.
• Counting loses credibility; the team stops believing the result and the count becomes a formality.

Catch it before it is produced

The answer is not counting more often. A count measures the past, and you cannot undo the past. What has to happen is that every legitimate way stock leaves gets recorded as it happens.

If waste is entered on a mobile terminal right where the item broke, it comes off stock together with its reason. If consumption is recorded when material is drawn for a job, it is clear which job it went to. Once those two records are in place, the count variance shrinks — and more importantly, whatever remains is a variance that genuinely needs explaining.

Counting then returns to its real job: not discovering loss, but confirming the system works.

Where to start

The fastest result usually comes from wherever waste and consumption actually occur: line-side in manufacturing, the packing area in distribution. Put a mobile terminal there and most unrecorded outflows become visible within the first month.

If you would like to work out where your own count variance comes from, we can look at it together with your own data in a demo.