There was once a product: a screen with a button. It had 1000 customers. On any day, 100 customers would click the button. Only in 80 cases, the button did the right job. The team came to know about the 20 failures. The engineer started thinking about a fix.

But the product manager thought of something else. He decided that they should introduce new button that did something else that was also equally interesting. The engineer claimed that it is also likely to fail 20% of the time and that it is a bad idea.

The new button got launched. Now there were 200 clicks going through each day, 100 per button. Only 70% went through without failure this time, which meant there were 140 correct executions each day.

The engineers were worried that their correctness is down from 80% to 70%. The company celebrated that their user engagement was up from 80 clicks to 140 clicks per day. More clicks meant more ad revenue. It called for a success party.