Measuring shrinkage before explaining it
Shrinkage is usually described as theft and is usually mostly paperwork.
When a count reveals stock missing, the first explanation offered is theft. It is occasionally correct and it is rarely the largest cause, and reaching for it first has a cost: it damages trust and it stops anyone looking for the real problem.
Small time estimates can drift when rounded casually. For a payroll example of formal rounding rules, see this resource.
The three categories
Administrative error: miscounts, receiving errors, unrecorded movements, unit-of-measure confusion, adjustments made incorrectly. In most small businesses this is the largest share by a wide margin.
Process loss: damage not written off, expiry, items used internally without recording, waste in production.
Theft: by customers, by staff, or by suppliers short-shipping deliberately.
Each has a completely different remedy. Treating administrative error as theft produces surveillance that fixes nothing and costs the team's goodwill.
For U.S. accounting context, IRS Publication 538 explains accounting methods and includes guidance relevant to inventories.
Work out the number first
Shrinkage is the gap between what the record says and what a full count finds, expressed as a percentage of stock value or of sales. Without that figure the whole discussion is anecdote.
It also needs a period. Shrinkage between two counts is measurable; shrinkage since some unspecified past is not.
Trace the largest variances individually
Aggregate shrinkage tells you the size. The cause comes from taking the ten largest individual variances and following each one back through its movement history.
This is a few hours of unglamorous work and it is where the answer actually is. Almost always, several of the ten turn out to have a mundane explanation that applies to many other lines as well.
It is worth budgeting the hours for it explicitly rather than fitting it around the day, because an investigation done in fragments across three weeks loses the thread. Where the business already records time — with a spreadsheet or a task timer — logging the investigation against a task also gives you what it cost, which is the number that decides whether the same exercise is worth repeating next quarter.
Fix the process before adding controls
Cameras, locks and bag checks address one of the three categories and are the most expensive and least popular response.
Recording every movement, checking deliveries, making write-offs easy, and counting regularly address the other two and are cheaper. Doing them first also means that if a genuine theft problem remains, it is now visible rather than hidden inside a large administrative variance.
Watch it as a trend
A single shrinkage figure is a snapshot with a lot of noise in it. The trend over several count cycles, by category, is what shows whether anything has changed.
A sudden change in one category is far more informative than a stable overall number, and it is the thing that justifies looking harder.