Theoretical vs actual stock: what the gap is actually telling you

Ask an operator how much beef they have and you will get a confident answer. Ask them to prove it and the confidence goes. The count says one thing, the POS implies another, and the invoice a third. Most people treat the disagreement as a data problem. It is not. It is the finding.

What theoretical stock really means

Theoretical stock is arithmetic, not a guess: opening balance, plus everything received, minus everything the recipes say was consumed, minus recorded transfers, waste and adjustments. Run honestly, it tells you what the shelf would hold if every process worked exactly as designed.

Actual stock is what someone counted. That is all. It carries no assumptions and no recipe theory — just a human, a shelf and a number.

The gap is a list of suspects, not a single cause

When the two disagree, the difference has to have come from somewhere, and there are only a handful of somewheres. Over-portioning at the line. Waste that was binned but never logged. A delivery short-shipped against an invoice nobody checked line by line. Theft. A recipe whose yield was set once and never revisited.

Each of those has a different fix, which is why a single variance percentage is close to useless. What you need is the variance attached to a product, a location and a date range, so the question stops being "why is food cost high" and becomes "why did prawns drift 1.8kg at the bar between Friday and Sunday".

Why the count has to be boring

Variance is only as trustworthy as the count underneath it, and counts collapse when they are hard. A count that takes three hours on a clipboard gets done monthly, and monthly variance is archaeology. A count that takes twenty minutes on a phone gets done weekly, and weekly variance is still actionable. The unglamorous conclusion is that the scanning matters more than the analytics.