A-B Split

The A-B split is a method of testing the effectiveness of marketing methods or media. Using A-B split marketing, a list of target names is split into two groups on a random basis, with one group designated as a control group and the other as a test or variation group. The objective of the A-B split is to determine which single variable is the most effective in improving response rates to a marketing campaign or achieving some other desired outcome. The A-B split is also referred to as "A/B testing," "bucket tests" or "split-run testing."

The A-B split has been been used for many years in direct mail campaigns, but has also successfully adapted for use in interactive media for testing the effectiveness of email blasts and banner advertisements.

Tests are divided into two groups: the control and the variation. For example, an email campaign by a newsletter publisher may include a specific "call to action" — such as subscribe within 48 hours to receive a 20% discount — embedded in the message to half the target audience, and no call to action (i.e. no solicitation to subscribe or mention of a discount) in the message to the other half of the test group. This will enable the publisher to determine if the "call to action" really works, and whether the response rate is good enough to justify the 20% discount.

While many parties present A-B split testing as reflecting a new way thinking about business strategy, it has been used in various research contexts for many years. The difference is how quickly and how frequently is can be deployed with the help of the internet. In fact, A-B split testing can be performed continuously at relatively little cost. Such testing allows for the real-time fine-tuning of marketing campaigns, website updates and development of online tools, which allows companies to keep abreast of changing trends. In effect, A-B split testing results in actionable guidance on website optimization that utilizes data in decision-making. There exists A-B split software that assists with such tasks. Some providers include Google Analytics, Optimizely, Unbounce, ion interactive and Instapage. 

The following steps can be followed to create and execute an A-B split test:

• Data collection: Data will reveal where a website or marketing effort should be optimized. Focus on both high-traffic areas as well as trouble areas.
• Identifying goals: Goals can range from a click-through, a conversion, email signups and more.
• Form a hypothesis: With a goal set, A-B split testing ideas can be formulated. They can then be prioritized by impact and implementation effort.
• Create test variations: A-B split variations can be subtle, such as changing the color of a key button, or broad.
• Run the test: This step deploys the test to users. Data is then collected from their interactions.
• Analyze the results: Data should be presented side-by-side to compare how they compare and whether the results are statistically significant enough to act on.