5 Basics of A/B Testing

Split URL testing can lead to huge benefits in your website or app. If you’re new to testing, you’re at the right place. These 5 guidelines are important to running any test.

Let your Tests Run Long Enough

Within minutes of letting your test run you might see huge gains in some of your variations. Don’t celebrate or panic, most likely a few conversions can throw off your results early on. Depending on what you are measuring, time required can vary. Luckily VisualWebsiteOptimizer made a handy test duration calculator.

Make BIG Changes

Why settle for a tiny gain in conversion when you could push the envelope and push your business to the next level? Things like colors, testing below the fold, and slight page modifications can only alter your conversion slightly. Doing smaller changes may be easy but alternate designs and testing big elements can lead to a huge gain – but does come along with more work.

Surprisingly the MS Paint ad had a nearly double click through rate

Don’t Assume

No matter how silly an idea may sound, everything is worth testing and can always lead to surprises. A great article on the Plenty of Fish blog shows an example of this, testing a professionally made ad versus an ad made in MS Paint. Surprisingly the MS Paint ad had a nearly double click through rate. It just shows that unexpected things can happen from trying unconventional ideas, and you’ll learn something every step of the way.

Make it a Real A/B Test

Running a variation of a page for a week, then running a new variation of that page is not a true way of running an A/B test. If you don’t run all variations of your page at the same time to a random selection of users you are getting skewed data.

Keep at it

The most important thing about testing is to keep testing even after having failing results.Learning about your customer and demographic can eventually lead to huge gains. There are plenty of resources for finding testing ideas, including the KISSmetrics blog and Unbounce blog.

The key tool I use in running A/B tests is VisualWebsiteOptimizer, after having countless problems with Google’s Website Optimizer I find paying for VWO is worth it.

What tools and resources do you use in your tests? Let us know in the comments