Experimentation has led to some of the greatest achievements in human history. A big part of that experimentation has stemmed from testing prototypes to see which was the best. That’s what A/B testing is. This is when you launch two similar products, or in this case websites, to see which gets better results. By creating a working hypothesis, and testing it with a side-by-side comparison, you can better understand which would be a more successful approach.

First, figure out what could improve. Review your current numbers so you can set a reasonable goal with how to improve them. For the sake of this post, we’ll pretend like we’re trying to increase subscribers to a newsletter (a common problem). You would want to review how many subscribers you have now, when they were acquired, and where on your website that they signed up from.

Next, set a reasonable goal. Generally, it’s better to set this as a percentage increase than a hard number. It’s also important to note that making big, overarching changes is hard to do without a number of factors coming together. To A/B test for success, it’s best to only test small sections at a time. For this exercise, we’ll call it a success if we increase our newsletter subscribers by 3%.

Once you have a goal, set a hypothesis and create the pages to match. Do you think it’s the placement of the newsletter registration that’s limiting subscribers? A/B test the location. Could it be the color of the subscribe button that’s getting lost in your layout? A/B test different button colors. Just be sure to keep the changes between websites small, and keep the testing live for enough time to receive valid results.

Last, you’ll analyze the numbers and see if you met your goals. This is the point in testing where you get to see if your hypothesis was correct. If it was, good for you! That means it’s time to do another round of A/B testing on another part of your website. If your hypothesis was incorrect, think about why and create a
new hypothesis.

User preferences change all the time. The best way to A/B test your website for success is to always be testing.