« How multivariate testing can change your whole business | Home | Doubling conversion rates: MarketingSherpa case study »
Test nothing and get results
By Billy | January 18, 2008
Now this is a blog about testing right? So why test nothing? Because nothing is powerful.
I don’t mean don’t test at all, I mean test having less on your page. Those awesome sub headlines your copywriter created? Or those testimonials? Your audience might not care or even look at them.
Best practices say to use trust logos, reviews, awards and a whole lot of other content, but you never know if it really helps. In fact, one of the biggest lifts ever at Widemile, came from removing everything but the core material on the page. One hero shot, one headline, one description and one button.
It was plain and simple.
To me and everyone else, it was an empty page filled with white space, but it converted at an extremely high rate compared to the previous page. Everyone at my office and at our client’s couldn’t believe it was the optimal page, but we couldn’t argue with the data.
While this might be an extreme example, it also is an example of the opportunities you may be missing if you don’t test turning off (hiding) elements on your page. So don’t just test something, test nothing too.
Related posts
Topics: Landing Page Optimization, Methodology, Testing Techniques |
January 28th, 2008 at 6:07 pm
Besides testing ‘nothing’ against ’something,’ another good test I think is useful every 6 months or so is to test the exact same page but labeled A and B. This way, you can ‘test’ your testing system. If one of the pages is a clear winner, then your testing environment sure isn’t!
January 29th, 2008 at 3:13 pm
That’s a great idea. You could also accomplish this with 2 very similar variables in a multivariate test, like “Buy today” and “Buy now” and if they performed drastically different, conclude the same thing.