ABC analysis in inventory management, adjusted for dated stock

ABC analysis sorts your catalog into three groups so that the handful of items carrying most of the value get most of the attention. The method is sixty years old, borrowed from manufacturing, and it works — right up until it is applied unmodified to a product with a date on it.

This page covers what ABC analysis is, how to run it, and the one change that has to be made before the output is safe to use for food and drink.

What is ABC analysis, and how do you define the classes?

ABC analysis ranks every item by annual consumption value — unit cost multiplied by annual usage — then splits the ranked list into three classes. A holds the small number of items carrying most of the value, C the large number carrying little, and B the middle. It is a way of deciding where to spend attention.

The underlying observation is that catalogs are lopsided. A brand with 200 SKUs will typically find that 30 of them carry two-thirds of the value, and that the remaining 170 are collectively worth less than the top 10. Treating all 200 with equal care means either over-managing the tail or under-managing the head, and in practice it means both.

Note what class C earns: generous buffers. That is counter-intuitive and it is the point. Cheap, slow items are not worth the management effort of a tight buffer, so the cheapest correct answer is to hold plenty and stop thinking about them. The scarce resource being allocated is attention, not stock.

What changes when the product has a date?

Class C stops being safe. The classic advice — hold generous buffers on low-value items so they need no attention — assumes stock waits patiently. Dated stock does not: a generous buffer on a slow C item is the single most reliable way to manufacture a write-off, because slow and over-bought is exactly the combination that expires.

The modification is to rank on value at risk rather than on consumption value alone. Value at risk asks not just what an item is worth, but how much of that value is exposed to the date: a cheap item with six weeks of shelf life and slow movement can be riskier per pound than an expensive one with a year. The ranking shifts, and several items that classic ABC files under C move up.

In practice that means running the classic ranking first, then applying a second pass that promotes any item whose cover exceeds its remaining selling window. Those items get C-level economic importance and A-level attention, which the classic method has no way to express — and it is the same measure that sets counting cadence by value at risk per counting hour.

How do you actually run it?

Multiply unit cost by annual usage for every SKU, sort descending, and cut the list where the cumulative value reaches roughly 70% and then 90%. Then apply the shelf-life pass. The whole exercise is one afternoon in a spreadsheet, and the mistake is treating the class boundaries as precise.

Two things keep it useful afterwards. Reclassify quarterly, because products move between classes and a stale classification is directing effort at what used to matter. And make sure something downstream actually differs by class — different counting frequency, different buffer policy, different review cadence. A classification that changes nothing is a tidy spreadsheet, not a method.

Where it usually pays first is the counting programme, because that is the clearest case of a fixed resource being spread evenly across items that do not deserve equal shares. Reclassifying and then leaving the count schedule flat is the most common way this exercise gets wasted. The same attention-allocation problem shows up across a CPG operation, wherever a fixed resource is spread evenly over uneven things.