also never attribute to malice what can be explained by hyper-optimization -- if all ad clicks were misclicks of this type, an AB test on a speedup might say to keep the jank version, without saying why
normally I'd say in-house users would notice any gmail issue + it must be intentional, but most in-house users won't get shown ads
if all ad clicks were misclicks of this type, an AB test on a speedup might say to keep the jank version
The problem is that misclicks are not actually super valuable for the ad purchaser. So if Google changed their ad behavior and your bounce rate shot up, you (or enough reasonable customers) would reduce your ad spend to compensate.
Hyper-optimizing for clicks would eventually be noticed.
Comments
also never attribute to malice what can be explained by hyper-optimization -- if all ad clicks were misclicks of this type, an AB test on a speedup might say to keep the jank version, without saying why
normally I'd say in-house users would notice any gmail issue + it must be intentional, but most in-house users won't get shown ads
The problem is that misclicks are not actually super valuable for the ad purchaser. So if Google changed their ad behavior and your bounce rate shot up, you (or enough reasonable customers) would reduce your ad spend to compensate.
Hyper-optimizing for clicks would eventually be noticed.
Hyper-optimizing for ad clicks instead of user experience is basically a synonym for malicious treatment of users.