A/B tests are likely involved because no human would suggest those changes in earnest, but what metric could those tests have improved!? Either it’s a series o

A/B tests are likely involved because no human would suggest those changes in earnest, but what metric could those tests have improved!? Either it’s a series of unchallenged false positives, or there’s a lot about web interface that neither I nor they understand, but I would challenge the idea that is the result of a well-informed A/B testing culture. It’s the output of a thousand monkeys running a thousand tests a day for three years, and releasing anything that hits the green line.