But on the internet, attention spans are extremely fragile. Web surfers are looking for immediate gratification. Do they really want to stick around and read a gazillion word sales letter? Does long copy outperform short copy, even on the internet?

I have a suspicion that no, long copy doesn't always outpull sorter copy on the web. And so I developed the following experiment for a client:

My client offers an online quiz to their site visitors. Their web page had relatively long copy explaining the quiz and a registration form that asked for the visitor's name and email address.

After measuring the traffic, we discovered that approximately 20% of visitors landing on that page registered and took the quiz. Admittedly, that's a commendable conversion rate, but since this landing page was available to only highly targeted traffic, I had a feeling we could generate an even higher conversion.

So I tested a new page with shorter copy and the same registration form. On this new, short copy page, 37% of visitors registered and took the quiz.

I decided to take the test one step further. I removed the registration form from the short copy page and replaced it with a simple "click here to take the quiz" button. Almost 73% of visitors who landed on this page took the quiz.

So, short copy with no user barrier appeared to be the winner. But take a look at what happened next...

Since the test results seemed to prove that the less copy the better, I removed all but one sentence of the copy and kept the headline and the button. No registration form, no barriers, no bulky copy getting in the way.

But only 55% of the visitors who landed on this page took the quiz. While shorter copy pulled better, there really is the risk of not using enough copy.

So how can you apply these test results to your own website? I would definitely suggest that you begin by testing shorter copy. And remove as many user barriers as possible. If the purpose of your web page is lead generation, you may not want to remove the registration form. Or, you may want to come up with a creative work around.

Time to implement: Basically, forever. That's because you should never be done testing your website. Once you've done an A/B test, take the winner and test it against something else. And then test that winner. And so on. With that in mind, you can expect it to take 2-3 hours to set up each individual test (depending on the testing platform you are using).