Also known as split testing, A/B testing is a way to test two ideas and analyze what gives the best result towards a pre-determined goal or a hypothesis in determining which ad would be successful. Basically, split testing enables a website owner to show two identical audiences the different version of some variation on a webpage at the same time. For example, you are running a print ad and you are not exactly sure on whether it would be better to show an image of your product or a person using your product. This is a perfect chance to do an A/B test to find out what could drive better results. To get started, let us continue using the print ad example. By implementing an A/B test, you would be able to take out un-important versions that will not deliver your key performance indicators (KPI, for short) but still you want to create different versions of an ad and likewise test them. To have a clearer perspective how to, you would need to ponder on the following:

1. Having a definite set of objectives – What are you trying to achieve with your ad? Is it to drive awareness, trials, or downloads – once you have an objective, you are now able to layout your plan.
2. What is your Hypothesis- Once that the objectives have been set, opinionated question would result; what do you think to perform the best? Would you be able to quickly take action based on the results of the test?
3. Create timelines and lists of what do you exactly want for your test – potential web elements for testing are colors, photos, graphics and texts such as headlines and the ad copy itself. Macro elements as well such as the ad concept, layout and number of visual elements.

There are a few important things to remember in doing accurate A/B tests to ensure fair results.

1. Test on variable at a time – make sure you are measuring the impact of one element change at a time. If version A has a different image and a different headline than version B, there would be a difficulty in pin-pointing which exactly could be causing the result.
2. Have the ad shown to similar audiences – If you are showing different versions of an ad to different audiences, chances are that your split test’s data will be totally inaccurate. It would be ideal to take your target audience and split the ad and measure both ads