Vilfredo Pareto was looking at who owned the land in Italy, found that a small minority owned most of it, and the same lopsided shape turned up everywhere anyone went looking afterwards. In a product it shows up as usage. A handful of things carry almost all of what people actually do, and the long remainder carries very little.
The part that gets quoted is the ratio. The part that decides whether the ratio is any use to you is which twenty per cent, and that is where the demonstration below spends its time.
There are twelve features. You have twenty units of effort and you are spending them before anybody has used anything. Put them where you think the use will be.
Then the product ships to 240 people. Each of them does between three and nine things, and each thing is drawn from a preference that falls off as one over the rank squared — the top feature is four times the second, nine times the third, and so on down. Which feature sits at which rank is shuffled after you have allocated. That is not a trick. It is the condition you were working under, made explicit.
Twenty units, twelve features, and no information. Where the effort goes is a guess until the thing has shipped.
The part that is only knowable afterwards
Whatever number came back for your effort on the top two, it is luck. The ranking was assigned to the feature names after your allocation was locked, so nothing you knew could have informed it. That is the honest version of the planning situation and it is why the second pass exists.
Press the button and allocate again with the ranking printed next to every name. It takes about four seconds and it lands twenty units out of twenty on the vital few. Nothing about the product changed between those two numbers. The only thing that changed is that the answer had already happened.
The bottom nine
The three features at the top are a quarter of the list and most of the volume. The other nine are the obvious cut, and the number that gets quoted for them is their share of use, which is small. The number that does not get quoted is underneath it in the panel: how many of the 240 people touched at least one of those nine. It is usually around half.
Those people did not spend most of their time down there. They went down there once, for the thing they came to do that day, and the export or the print layout was the reason the session was worth having. Volume and reach are different measurements and only one of them is in the slogan.
What it costs you
The ratio is a description, not a plan. Eighty-twenty is what the data looked like after the fact. It contains no instruction about what to build next, because the ranking that makes it true does not exist until people have used the thing.
Concentration in volume is not concentration in value. A feature used once by 90 people out of 240 is a rounding error in the usage chart and a reason 90 people stayed. The chart does not know the difference and will not tell you.
Beware of measuring the ranking on the users you already have. They are the ones the current shape suits. The features they barely touch are also the features that nobody who needed them stayed long enough to touch.
The vital few move. A ranking taken this quarter is a fact about this quarter. Ship a change, and the ranking you cut against is describing a product that no longer exists.
Run it a few times. Some runs come out at eighty per cent for the top two and some come out at sixty-something, off the same fixed preference, because 240 people doing between three and nine things is not very many. The slogan is a shape that turns up often enough to be worth knowing about, and the arithmetic underneath it was never a law.