Ask a plant manager how much scrap they ran last month and you will usually get a good answer. Ask the general ledger the same question and you will often get a different one. The gap between those two numbers is where manufacturing costing quietly stops being useful.
The cause is almost never the accounting. It is where and when the scrap is recorded.
The common pattern: an operator notes scrap on a sheet at the machine. The sheet goes to a supervisor. At month end someone keys a total into the system, or posts a single adjusting entry. It works, in the sense that the inventory balance ends up roughly right.
What it destroys is attribution. One number at month end cannot tell you which work order, which operation, which shift or which lot the loss came from. So every downstream question, is this a tooling problem, a material problem, a training problem, has to be answered from memory rather than data.
It also lands the entire variance in one period, on one day, with no warning. The first anyone in finance hears of a bad month is after it closed.
Scrap should be recorded at the operation that produced it, by the person who saw it, at the time it happened, against the work order it belongs to. In NetSuite terms that means it is entered as part of work order completion rather than as a separate inventory adjustment later.
The immediate change is that scrap stops being a month-end surprise and becomes a running number. Beyond that, three things become answerable that were not before.
The third row is the one with money attached. If scrap is not attributed to the operation, standard cost carries an optimistic yield assumption, and every price built on that cost is quietly too low. On a high-volume, thin-margin part that is not a rounding error.
Operators will say this is more data entry, and they are right, it is. The trade is worth making only if the data comes back to them. If the scrap they key in appears on a shift board the next morning and leads to a fixture being replaced, capture holds. If it disappears into a finance report nobody on the floor ever sees, it degrades within a month, and no amount of system design prevents that.
Not with the full scheme. Pick the one part number with the worst suspected yield, capture properly on that line only, and run it for a month. You get a real measurement, the operators get a short list of reasons rather than an abstract policy, and you find out what the capture actually costs in seconds per completion before committing the whole plant to it.
Almost every rollout we have seen that started plant-wide ended up back at a paper sheet. The ones that started with a single line are still running.
Six questions about your business, and an indicative phase plan with the arithmetic shown.
Years of building on NetSuite: SuiteScript, SuiteFlow, SuiteQL, SuiteTalk, Advanced PDF templates and Celigo integrations. We write these up when a problem comes round for the third time.
Tell us what you are running today and where it is breaking down. Most calls run 20 to 30 minutes and end with a clear next step. We reply within one business day.