The 5 Steps of Successful Split-testing
By Rachel Anderson
So you want to test ads? Great! Don’t know where to begin? I’ve got you covered. Remember that split-testing, or A/B testing as it’s also known, is essentially creating 2 versions of an ad to discover which achieves a determined goal more successfully. ‘A’ is usually the existing ad, called the control, and ‘B’ is the variation of it. You divide traffic evenly between the 2 and see which performs the best.
Here are the 5 steps to conduct a winning spilt-test:
1. Define Success
Before you doing anything else, set a clear objective for the test. What’s the goal?
It could be a number of things including but not limited to:
- Boosting traffic
- Driving sales
- Increasing sign-ups
- Lowering bounce rate
Whatever you choose, make sure it’s well defined and attainable.
2. Decide How to Test
Although the goals of spilt-tests greatly vary, there are a surprisingly good number of standard ways to perform the test. Here are a few:
- Use separate landing pages
- Different calls to action
- Unique headlines
- Caps verses no caps
- Including price verses no price
- # of fields in a form
- Length of text
- Different landing page images
3. Run Ads Simultaneously
It’s extremely important to the accuracy of your split-test that you start running your ads at the same time. I’ve heard of people thinking you should wait to test the variation until you’ve first tested the control. This will dramatically skew the data. The whole point of a split test is to see which ad achieves the goal better. If there is only one option at a time you will never really know. It all hinges on both ads running at the same time for the same amount of time.
4. Don’t End the Test Early
You may be tempted, for a number of reasons, to end the test early. Don’t! Looking at the data you may think you already have a clear winner but wait it out. Set an amount of data to collect before you even start the test and then stick to meeting that requirement. You may be surprised at the result!
5. Use Results to Begin Your Next Split-test
Once you finish your split-test, don’t stop there! You’re first test may have been a flop but don’t be discouraged. Try testing another way. If your test was a success, you’ve learned that something works better than another. Use the winning ad as the new control and conduct another test. See if you can find something that works even better. It’s a simple and pretty fun way to continuously improve how well you achieve account goals.
Rachel Anderson is a Pay Per Click Advertising Strategist at Netmark.com. Off the job she enjoys photography, good food, being outside, and spending time with her husband. Share your thoughts with Rachel on Twitter @gladygirl, Google+ or Facebook- she’d love to hear from you!