Tales from Silicon Valley #1: The Fucking 80/20 Rule

me irl

What is the 80/20 Rule?

The 80/20 rule is the idea that in the total population of Thing Y, 80% of Action X is completed by a 20% segment of Thing Y.

The idea that at any given company 80% of the work is completed by 20% of the people is the most common example of the 80/20 rule. This is either uplifting or disheartening depending on which group of employees you assign yourself to. The 80/20 rule we would use the most in my line of work was to assert that 80% of failures of an engine,tree or cow would be caused by the top 20th percentile of most common failure types. These both sound true but they may also not be true — I’m not a data scientist so I have no idea.

The beautiful thing about the 80/20 rule is that it can be used to justify whatever your particular obsession is. 80% percent of violent crimes are committed by the top 20% of reoffenders, so increase mandatory sentencing. 80% of airline crashes occur on flights run by the 20% of airlines that have the most safety violations. 80% of fatal car crashes happen on the 20% of drives that make up short trips around the neighborhood, so always buckle up on short trips. 80% of our identity is constructed in the first 20% of our lives so make sure to be a good parent.

The 80/20 rule has a lot of great uses, one of which stands out above all others. The 80/20 rule can be used to justify doing a half-arsed job. More than that, it can be used to justify doing a 20% arsed job. You can only do two half arsed jobs at once. You can do five 20% arsed jobs. That’s a 250% increase in arse productivity!

It’s nice because it sounds smart

Now it might sound like I’ve just regurgitated a bunch of semi-useful platitudes dressed up in the language of data science. And that’s exactly what I did. But if I told you I was a data scientist, would you listen to me? Probably.

In my experience it usually goes something like this:

The data scientist is brought before the people and she gazes at the data. The data scientist closes her eyes and says “80% of your problems are caused by the 20% of your assets that are the worst performing.” There is a pause. People start nodding. People start clapping. The wisdom is self evident. They way forward is now clear.

I’m not jealous. Which brings me to the story-

Winning friends and influencing people with data science

Many years ago I was in a non-english speaking company selling software for a company that’s definitely not on my LinkedIn profile so don’t even try looking for it. There was me, the Salesman and the Data Scientist. We were selling a new product we were working on to a local company. The company was quite old fashioned and didn’t contain a single soul that spoke english. The salesman was a local and fluent in both languages so he was acting as the translator.

We were meeting with middle managers who would be using the software. The meeting was going quite well. The Salesman, the Data Scientist and I explained all the problems that were solved and the opportunities that would be provided by using our software. They seemed quite impressed. Our software was quite new however and one of the middle managers correctly pointed out that it wouldn’t be able to support 100% of their issues right off the bat. This was an astute observation and a problem inherent to adopting any new software system. At this point the Data Scientist jumped in and explained the Mighty 80/20 Rule, which the Salesman translated.

There was a pause. The middle managers started nodding. They made impressed noises. The Salesman said to the Data Scientist:

“Could you explain the 80/20 rule in more detail?”

The Data Scientist, for his part, did not particularly enjoy explaining the 80/20 rule. But he started talking about silt, airplane engines, probability and all the awesome stuff that makes data scientists so great and terrible. The middle managers were very impressed. The meeting ended on a hopeful note.

The next day, we met with another set of middle managers. 5 minutes into the meeting the Salesman asked the Data Scientist to explain the 80/20 rule to the new set of middle managers. The Data Scientist rolled his eyes a little but complied. These new middle managers made even more impressed-sounding noises than the original set of middle managers. The 80/20 rule was a hit. It was a big hit. It was the killer single that would carry our album to mega-platinum.

On the final day of our customer visit, we were in a taxi on our way to meet the senior people of the company. The director level people. The Salesman told the Data Scientist that he should be ready to explain the 80/20 rule again in this meting. The Data Scientist said that he didn’t want to explain the 80/20 rule again, because he didn’t feel that it was relevant. The Salesman said that may be true, but it has gone over so well with the other teams that it would be foolish not to play our best card. They agreed to disagree.

As you may have expected, once we got into this final meeting the Salesman again asked the Data Scientist to explain the 80/20 rule. This time the Data Scientist ignored the request. He went on to talk about his true passions: silt, engine fires, magnesium parts per million and why you shouldn’t fly planes through atmosphere that contains 100 magnesium parts per million.

The Salesman paused for a moment and said:

“そつ 80/20 Rule ささひ しすふふ ふいはき しくし はく あふお えうまはきと. 80/20 Rule いはき しくし はく.”

The director guy replied:

“80/20 Rule いはき しくし はく?”

The Salesman said:

“いはき しくし はく80/20 Rule.” Then he turned to the data scientist and smiled. The 80/20 rule would be presented and translated from the Data Scientist whether he wanted it to be or not.