ABC Analysis: Classify on Two Dimensions, Not One
SRF Capital Studio Research DeskFunding Intelligence, SRF Capital StudioA classification that assigns different management policies to items that deserve them, and the trap that turns it into an expensive mistake.
Summary
- ABC analysis ranks items by contribution and splits them into classes, so that a business can stop applying one policy to a population that is nothing like uniform.
- Rank on contribution, never on revenue. A high-revenue line that earns nothing will otherwise be managed as though it were the most important thing in the range.
- The common failure is deleting C items on their own numbers. Customers buy ranges, and a small line can be the reason a large order exists at all.
What an ABC classification actually does
ABC analysis takes a population of items and sorts it by contribution, then splits the sorted list into classes that each get a different management policy. It applies to stock-keeping units, customers, suppliers, inventory lines, or anything else a business holds a lot of.
The idea underneath it is the Pareto observation that value in most populations is heavily concentrated. A small share of items carries most of the contribution, and a long tail carries very little. What ABC adds is the operational consequence: those groups should not be managed the same way.
A items get attention, tight control and short review cycles. C items get rules, low touch and as little management time as the business can get away with. The point is not the classification. It is the differentiated policy that follows it.
Rank on contribution, never on revenue
This is the most common error and the easiest to fix. Ranking by revenue promotes high-turnover, low-margin lines into the A class, where they absorb attention out of proportion to what they earn.
- Choose the ranking metric. Contribution, not revenue.
- Rank and compute cumulative share. Sort descending, then accumulate down the list.
- Split the classes. A is roughly the top 70 to 80% of value, B the next 15%, C the remainder.
- Assign a policy per class. Service level, review frequency, stocking rule, order minimum.
- Re-run it annually. A classification built three years ago is still driving today's stocking policy, and the population has moved underneath it.
The class boundaries are conventions rather than laws. What matters is that the cut points are defensible and that they produce groups a business can genuinely treat differently. Three classes that all receive the same handling have cost a week and changed nothing.
Value alone gives you a weak classification
The textbook version classifies on one dimension. It is the weakest useful version of the method, and the fix costs almost nothing.
Classify on value against volatility instead. A high-value item with stable, predictable demand needs a completely different stocking and service treatment from a high-value item whose demand arrives in unpredictable bursts. The first can run lean. The second needs buffer, or it will fail a customer at the worst moment.
Building the two-by-two costs no extra time, and it changes the policy recommendations materially.
Once the second axis is there, the policies almost write themselves. Stable and valuable means tight control and low safety stock. Erratic and valuable means tight control and real buffer. Stable and small means a rule and no meetings. Erratic and small is where a business should be asking whether to carry the item at all, which brings us to the trap.
The basket effect that destroys value quietly
Customers buy ranges, not line items. This is the single most expensive thing to forget when a rationalisation exercise reaches its conclusion.
A C-class item, judged on its own , looks like an obvious deletion. It may also be the reason an A-class customer places the order at all. Remove it and the customer consolidates their purchasing somewhere that carries the full range, and a business that set out to improve its margin loses the basket instead.
The check is not complicated. Before recommending any C-class deletion, look at which customers buy the item and what else they buy at the same time. If the answer is that the line appears in orders from the most valuable customers, it is not a C item in any sense that matters. It is a range item with a small number attached.
This is also why customer-level work and item-level work belong together. Which customers actually make money is a different question from which products do, and the metrics that carry a diagnosis covers how the two connect.
Policies that should differ by class
- Service level. Availability commitments that are worth funding for A items and not for C items.
- Review frequency. Weekly attention on A, quarterly on B, exception-only on C.
- Stocking rule. Where safety stock is held, and where the business is content to be out for a week.
- Order minimum. The threshold below which a C-class line is not worth picking, packing and invoicing.
- Authority. Who can discount, and by how much, without a second signature.
Cost to serve belongs in this conversation as well. An item's contribution as reported rarely carries the cost of handling it, and small erratic lines absorb warehouse and administrative effort far out of proportion to their value.
The policy differences are also what make the exercise pay for itself. A classification that produces no change in how anything is ordered, stocked or reviewed has cost a week of analyst time and delivered a spreadsheet. The test of a finished ABC is whether someone's standing instructions changed as a result of it.
In mid-market Indian businesses the constraint is often not knowing the right answer but having the system to apply it. A differentiated service level needs something in the order-taking process that recognises the class. Where that runs on a person's memory rather than on a rule, the policy decays back to uniform within a quarter. That is a process question rather than an analytical one.
Which is why the classification is worth rebuilding on contribution the business actually believes. Where allocated overhead does most of the work in the ranking, the argument shifts. It stops being about what to do with the C class. It becomes a question about whether the ranking itself is sound. How to benchmark a business covers the normalisation that settles it.
Re-run it, or last year runs your warehouse
A classification is a snapshot of a population that keeps moving. Items are added, demand shifts, a competitor exits and a quiet line becomes a busy one.
Re-running it annually is the minimum. The failure is not that the classification becomes wrong in some abstract sense. It is that a policy set on old data keeps allocating stock, service and attention as though nothing had changed. Nobody notices, because the system still produces a confident-looking output every month.
The tell is usually a complaint rather than a report. A salesperson says a line is always out of stock, or a warehouse manager says nobody ever asks for something that takes up a full rack. Both are describing a classification that stopped matching the business, and both are cheaper to act on than the annual review that should have caught it.
It is worth being clear about what the method does not do. It is purely descriptive. It ignores strategic necessity, such as range completeness, regulatory lines, or an item carried specifically to block a competitor. And it ignores interdependence, which is the basket trap again. Those judgements stay with the people who know the business, and a classification is an input to them rather than a substitute for them.
Frequently asked questions
What is the difference between ABC analysis and Pareto analysis?
Pareto is the observation that value concentrates in a small share of items. ABC is what a business does about it: split the ranked population into classes and give each one its own policy. Pareto describes, ABC operationalises.
What percentages define A, B and C classes?
A is conventionally the top 70 to 80% of value, B the next 15%, and C whatever remains. Treat those as starting points. The right cut is the one that produces groups the business can actually manage differently.
Can ABC analysis be used for customers as well as inventory?
Yes, and it is a fast first cut before heavier customer profitability work. The basket caution applies with even more force there: a small customer may be a subsidiary of a large one, or the route into a group relationship.
How often should the classification be rebuilt?
Annually as standard, and sooner after anything that reshapes demand, such as a range change, a large customer win or loss, or a competitor leaving the market.
How useful was this article?
One tap. It tells us what to write more of.
About the author
SRF Capital Studio Research Desk
Funding Intelligence, SRF Capital Studio
The next one
Get what we publish next, by email.
Working notes on raising, borrowing, protecting, growing and structuring capital in India. One email a week at most, and you can leave any time.
We use your address only to send this. See our privacy policy.
We store your address to send you these emails and nothing else. See our privacy policy.
Related reading
