How to do an ABC analysis
Who it is forGrocery chain buying, category management, inventory planning
In a catalogue of a thousand lines you cannot give every product the same attention, and you do not need to. A small share of products produces most of the value; the remaining majority, however numerous, has limited total impact. ABC analysis makes that split. XYZ adds a second dimension: is this product's demand predictable. Together they make it clear which inventory policy each group needs.
How to do an ABC analysis
ABC analysis rests on the Pareto principle: most of the value comes from a minority of products. The analysis takes five steps and can be completed in a spreadsheet in half an hour.
The critical decision is in the first step: what base to rank on. The common mistake is ranking by units sold, which puts cheap high-volume items into A class and expensive low-volume items into C. The correct base is annual consumption value: units multiplied by unit cost.
Different decisions can use different bases. If you are managing inventory investment, use consumption value; if you are managing shelf space, use volume; if you are managing profitability, use gross profit. When the base changes the classes change, and that is normal.
- Choose and calculate the baseFor each product, annual consumption value = annual units sold × unit cost. The period must be at least twelve months; short periods mislead for seasonal products.
- Rank from largest to smallestProducts are ordered by descending consumption value.
- Calculate cumulative percentageEach row shows the running total of value up to that product as a share of the grand total.
- Apply the class boundariesCumulative up to 80 per cent is A, 80-95 is B, 95-100 is C. The boundaries are not sacred; 70/90 also works depending on your distribution.
- Write a policy per classCount frequency, service level, ordering method and approval level differ by class. A classification that produces no policy is just a list.
| Class | Value share | Product share | Count frequency | Service level | Ordering method |
|---|---|---|---|---|---|
| A | 80% | 15% - 25% | Monthly | 97% - 99% | Continuous review, reorder point |
| B | 15% | 25% - 35% | Quarterly | 95% | Fixed-period ordering |
| C | 5% | 45% - 60% | Annually | 90% | Bulk orders, wide intervals |
The first 20 per cent of products produce 80 per cent of the value, and the curve flattens sharply after that: the 50 per cent of products sitting in class C add only 5 per cent to the total. Where to place the class boundaries is read from the bend in this curve.
C class is numerous and it is where the management cost sits. The goal in C class is not optimisation but simplification: reduce order frequency, increase batch size, and stop spending time on it.
Which criterion should you classify on
The ranking base is the single decision that determines the outcome of an ABC analysis, and in most implementations it is chosen without discussion. A different base produces different classes; both are correct, because they answer different questions.
The table below gives six common criteria, the question each serves and where each misleads. You are not obliged to work with one: a serious setup combines two or three criteria with weightings.
In a multi-criteria ABC, each criterion is converted to a percentile rank per product, multiplied by its weight and summed, and the classification runs on that combined score. A common weighting is consumption value 50 per cent, gross profit 30 and order lines 20.
| Criterion | What it measures | Which decision | Where it misleads |
|---|---|---|---|
| Annual consumption value | Capital tied up | Stock investment, ordering policy | Pushes cheap but critical items into C |
| Gross profit contribution | Profitability | Category and assortment decisions | Understates low-margin traffic drivers |
| Units sold | Operational load | Warehouse, handling, shelf plan | Pushes expensive slow movers into C |
| Number of order lines | Frequency of customer contact | Service level targets | Brings low-volume items forward |
| Number of customers | Strategic dependency | Identifying critical items | Demotes large single-customer items |
| Volume or shelf space | Space constraint | Grocery aisle and shelf management | Understates small high-value items |
Start with one criterion. Multi-criteria ABC is more accurate, but the analysis stalls for months while the weightings are debated; a classification built on annual consumption value always beats one never built at all.
An ABC classification worked through
To keep the analysis concrete, work through a small catalogue of ten products. Rank them by annual consumption value from largest to smallest, calculate the cumulative percentage and apply the class boundaries.
Because the example catalogue has only ten products, A class covers 40 per cent of them. In a real catalogue A class is a far smaller share; the class boundary is not a rule but a threshold set from the catalogue's distribution.
Keeping the analysis alive
The most common fate of an ABC analysis is being done once and forgotten. Products change class: a new product climbs from C to A, one whose demand fades falls from A to C. Stock managed against last year's classification is not this year's stock.
The right cadence is quarterly. At the end of each quarter the analysis is re-run and a list of products that changed class is produced. The real information is not in the classes themselves but in the movers: a product climbing from C to A is a growing opportunity, one falling from A to C is an eroding category.
New products should be kept out of the analysis. A product with less than three months of history is not classified correctly by ABC; those products get a separate "new product" policy and join the matrix once they have history.
Frequently asked questions
- Can the ABC class boundaries be something other than 80 / 15 / 5?
- Yes. The boundaries are not a rule but a threshold chosen from the catalogue's distribution: if 80 per cent of value sits in 5 per cent of products, narrow the A class; if the distribution is flat, widen it. What matters is that the boundary is written down and reused each quarter; if it moves, the list of products that changed class loses its meaning.
- How is an ABC analysis done?
- Products are ranked by annual consumption value from largest to smallest, a cumulative percentage is calculated and classes are assigned: the first 80 per cent is A, 80-95 is B, 95-100 is C. The ranking base is not units but units multiplied by unit cost.
- What are the ABC analysis percentages?
- By value, A is 80 per cent, B is 15 and C is 5. By product count the split is typically A 15-25 per cent, B 25-35 and C 45-60. These boundaries are a starting point rather than a rule and can be adjusted to your distribution.
- Which base should an ABC analysis use?
- It depends on the decision. Annual consumption value for managing inventory investment, volume for shelf space, gross profit for profitability. Ranking by units sold is a common mistake that pushes cheap items into A class.
- How often should an ABC analysis be repeated?
- Quarterly. Products change class, and the real information is in the movers: a product climbing from C to A is a growing opportunity, one falling from A to C an eroding category. Products with less than three months of history are left out.