ABC curve and Pareto classification of items
Your data
The value can be whatever you are ranking by: money moved in the year, money sitting in stock, hours of machine time. Change the criterion and the classes change with it, which is the point rather than a flaw.
The cutoffs are fields because eighty and ninety-five are a convention, not a law. If your list is unusual the convention will misplace things, and moving the line is cheaper than pretending it fits.
Results
| Items counted | — |
| Total value | — |
| Share of the value held by the costliest fifth of the items, in percent | — |
What each class ended up holding
| ABC class | Number of items | Share of the items | Value in the class | Share of the value |
|---|
The list, sorted
| Position | Item name | Value of the item | Share of the total | Running total | ABC class |
|---|
The dotted diagonal on the chart is what a list of identical items would look like. The further your curve bends above it, the more the classification is telling you something you did not already know. A curve lying on the diagonal is the honest signal that this method has nothing to offer for that particular list.
Two lists of the same items in a different order give the same classes here. Items worth exactly the same are ordered by name rather than by the order you pasted them, because a classification that moved when you shuffled your spreadsheet would deserve the suspicion it got.
The item that crosses the line still belongs to the class above it. Cutting before it would leave out the very item that completes the share you asked for, which is why a class A of eighty usually ends up holding a little more than eighty.
Is the eighty-twenty rule always true?
No. It is a common shape, not a law, and this page measures your list instead of assuming it. If the costliest fifth of your items carries 62 in a hundred of the value, the page says 62.
The number worth looking at first is exactly that one. A list of identical items would put 20 in a hundred in its costliest fifth, so the further you are above twenty, the more the classification is telling you something.
| Costliest fifth carries | What it means |
|---|---|
| About 20% | flat list, classes buy nothing |
| About 50% | mild slope, classes help a little |
| 80% or more | steep, the classic case |
When is an ABC classification not worth doing?
When the curve is nearly straight. If every item is worth about the same, splitting the list into three classes gives you three lists to maintain and no new information, because there is no small group of items that deserves different treatment.
That is a real answer and almost nobody measures it, because the rule is usually repeated as though it applied everywhere. The page prints the verdict in plain words above the tables.
Why does class A end up holding more than eighty percent?
Because the item that crosses the line still belongs to the class above it. If the running total reaches 76 and the next item takes it to 83, that item is what completes the share you asked for, so it goes into A.
Cutting before it would leave class A holding less than the eighty you asked for, which is the opposite of what the cutoff means. The overshoot is small on a long list and larger on a short one.
Does the order I paste the list in change the result?
No. Items are sorted by value, and items worth exactly the same are then ordered by name rather than by the order they arrived in.
Without that second rule, pasting the same spreadsheet sorted differently could move an item from B to C with no change to the data, and a classification that moves when you shuffle your input deserves the suspicion it would get.
Reactions
0
0 Comments
Be the first to comment