[heldtogether]

I still cite this article when discussing A/B testing with clients. They often want to bypass the testing process because "it's an obvious change".

"There's no point in making trivial adjustments, ... on a site this big and this complex, it is impossible to predict how even the smallest changes might affect the bottom line. Fixing the wonky images, for instance, might actually hurt Plenty of Fish. Right now, users are compelled to click on people's profiles in order to get to the next screen and view proper headshots."

This isn't to say that everything must be A/B tested, but it's very easy to overlook the fact you can negatively affect the site by making an apparently positive change.

[sytelus]

This guy is essentially saying that providing better customer experience would reduce number of clicks and his pageviews and thus his income. I think almost nothing should prevent providing better customer experience. This is also a problem with A/B testing. If your test is optimizing page views or ad revenue as opposed to customer satisfaction, you are probably doing it wrong. Better matrix might be increase in number of unique users or return rate or user churn rate or session abandonments rate etc.

[darkmighty]

You're being too dogmatic about "user experience". If users find more matches on the site, I'd say their experience improved.

Design principles such as 'the user shouldn't find anything difficult', 'things should be as visually pleasing as possible', etc. can be inconsistent sometimes. A better principle is just looking at whether the user goals are being achieved as effectively as possible.

[downandout]

>This guy is essentially saying that providing better customer experience would reduce number of clicks and his pageviews and thus his income

The problem is that "better" is in the eye of the beholder. That is why split testing is favorable even if the proposed changes seem obviously "better". There is no legitimate reason to skip split testing, but there are a number of reasons to not skip it.

[loceng]

Knowing what you're optimizing for and what your product value is are important. I have seen some highly successful viral platforms destroy themselves with a redesign because they didn't seem to understand the importance of their UI/UX - they thought their idea on its own was great.