We are launching a blog about web design. The goal of our blog is of an ordinary content site - to get visitors to read as many articles as possible (NOT to get ad impressions).

Our designer has produced 2 variations of the homepage, with the only difference being the shape of blog post items.

None of them is significantly better in our eyes, so we decided run an A/B testing.

My question is - What metric should we measure during the A/B test in order to choose the better design?

My instinct is Bounce Rate, but I suspect there's a better answer.

  • Related - (possible duplicate actually) ux.stackexchange.com/questions/8536/…
    – JonW
    Dec 17, 2012 at 16:22
  • 1
    Do you even have enough visitors for an A/B test to meet statistical significance? I think you've got the cart before the horse here... Dec 17, 2012 at 16:28
  • We're not doing the A/B testing now. We'll wait until we have a substantial amount of content and traffic to test on. Dec 17, 2012 at 16:32
  • What do you mean by the shape of the blog post items? Dec 17, 2012 at 18:32
  • The homepage has a list of posts. The difference in the design is the shape of the post items. Each item has an image. In one design those images are rectangular and in the other they're circular. Dec 18, 2012 at 8:28

2 Answers 2


There's no one metric to A/B test anything really. Conversion rate may be the best metric for ads/sales pitches, but even that's often an oversimplification.

Instead, look at all your metrics and see what needs work. Do you have a concerning bounce rate? Is time on page low? Are users not clicking the links you want them to click? Trying to improve an already low bounce rate is a waste of your time, and might be extremely frustrating as you find there really isn't anything wrong.

Normally you should create designs based on whatever you feel needs improvement, but since you already have both designs, I'd just A/B test both of them and keep an eye on as many factors as you can; bounce rate, time on front page/articles, which links get clicked, anything you can. Judge the two designs holistically. A Site with low bounce rate on the first page but high bounces on the second level (because users were misled/confused) could actually be an indication of a worse design than a landing page with high bounce rate (because people, correctly, realize they're legitimately not interested in the content, which is clearly defined).


I'd have a look at heat maps. A good few analytics providers appear to be offering them these days.


I've always found they are excellent for gauging a user's level of interaction with a content heavy pages.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.