5 Tips for Effective A/B Testing

July 19, 2013 Jason Corrigan

If content is king in the world of digital marketing and organic search, then A/B testing is the prince of a successful digital marketing campaign. The justification for this claim comes directly from Google’s guide for A/B testing, which contains over 230,000 results – obviously an impressive amount.

Businesses and brands who already develop organic search and digital marketing campaigns are most likely aware of the value of A/B testing, but the question that most businesses run into is: “What now?”

Well, the answer is to focus on different components of your campaign and reallocate effort and resources to support and fulfill secondary and tertiary aspects of your professional website.

Brands can find out and understand what aspects of their campaign and efforts are succeeding by creating experiments that measure the impact of a particular action, such as adding a new landing page to your site or manipulating user navigation.

Check out these five great tips that allow businesses and brands to avoid mistakes when experimenting with A/B testing.

1. Sample From A Large Pool

Too often, businesses and brands will end their testing cycle as soon as something comes out as a win or a loss. Marketers and executives get too excited or discouraged after one or two tests that they often run with the information they have gained. However, this can lead to critical mistakes later on as insubstantial information can lead you astray.

Therefore, making sure that you run a significant amount of tests is critical to truly understanding what does and does not work for your website.

For example, if you publish an article on your website and it receives 10,000 views over the next two weeks; does that mean you can expect a certain percentage of social shares? Of course not!

However, if you publish 10 articles over the course of 10 days and each receives 5,000 views and 400 social shares, then you are able to make an accurate hypothesis based on recurring data.

2. Period Of Testing

It is critical to remember that every website and digital marketing campaign is unique. Just because someone is able to get on page one of Google search listings by having a substantial blog, doesn’t mean that your website necessarily will if you copy their efforts.

Outside variables such as offline marketing, quality of links earned, and even sales cycles considerably affect search listings.

Therefore, you want to perform tests that have a significant amount of time behind them, so that you are able to obtain a true average in your results to compare and learn from. Depending on your campaign goals and industry, the duration you test for may differ. However, everyone should run as many tests for as long as possible in order to receive true and complete results.

3. Get Rid Of Irrelevant Data

While it is easy to get caught up with how much data can be pulled and learned from, it is important that you don’t let yourself get carried away with data that may be useless to your overall development.

When running a particular test, make sure you are mindful of certain days throughout the year that may have a dramatic impact on your results. Holidays such as Memorial Day Weekend or Black Friday are notorious for draining and spiking traffic levels. So when you’re performing a test, make sure you exclude these types of dates from your test.

For example, if you are interested in testing your website’s call-to-actions for a particular product or service, excluding days such as Easter Sunday will allow you to obtain much more accurate results since fewer people are likely to be browsing the internet during this time.

Beyond holidays, make sure you are aware of all media outlets that pick up and publicize your company. Having your name in the news can cause a dramatic spike in traffic so be considerate of such times so that you have the most accurate data possible.

4. Test Over And Over Again

Similar to making sure that you are testing with a large pool of samples, you want to make sure that you perform multiple rounds of testing. In addition, make sure that you frequently monitor your data for consistency throughout an experiment.

Make sure that a control code is placed correctly and that deviations from past performances are not hindering your current efforts.

5. Test By Funnel Stage

Ensuring that you are constantly focusing your efforts on the same stage of the funnel is essential in understanding consumer behavior and habits. Make sure that when you test for consumer behavior on your website, you are inserting the proper tracking codes that monitor what consumers do and where they go on your site.

Worth A Closer Look

It’s important to remember that the results of your tests are really only as strong as the experiments you have put in place. If you are not testing in multiple environments and testing continuously throughout your search and digital marketing campaigns, then you are not going to have credible results to base your next moves on.

A/B testing is an enormous aspect of a quality campaign and taking the time to set up proper tests will only help you grow your business by adapting to what consumers and search engines are looking for the most.

Got any more tips for effective A/B testing? Let us know in the comments!

About the Author

Jason Corrigan is an SEO Manager at Wpromote in El Segundo, CA who oversees the planning and execution of custom SEO and social strategies for medium sized businesses to Fortune 500 companies. He is a published author on Social SEO and has extensive marketing experience that ranges from journalism and content development to TV and Radio marketing/advertising.

Previous Article
Meta Elements: Small but Essential, or just Small?
Meta Elements: Small but Essential, or just Small?

When it comes to understanding data and search results, the needs of human users...

Next Article
How to Prepare your PDF for Uberflip
How to Prepare your PDF for Uberflip

One question we get quite frequently here at Uberflip is from people wondering how...

Check Out Our Upcoming Webinars and Events

Find Out More