ReTargeter Blog

How to Develop a Testing Strategy That Works for Your Business

Designing a company website is important. As the front line to the public and the first impression many potential customers will have to your brand, a website is a representation of your brand’s voice and personality; put simply, it matters.

However, during the initial design phase, or later revamps or makeovers, small changes can make a big difference in terms of conversions, traffic, search rankings, and overall success. Because of this, split testing, or randomized experiments that predict the success of specific messages, layouts, URLS, and more, are critical. So how can you develop a testing strategy that works for your business?

While researching testing strategies, two major forms will become apparent: A/B testing and multivariate testing. These research methods each have their benefits and inherent limitations. To pick the method that will be most effective for your brand, it’s important to take a comprehensive look at each method.

The Importance of Testing

Web design can’t be a haphazard endeavor where information is simply broadcast to the world in hopes that it will be successful. That’s not how business works in the real world, so it’s certainly not acceptable online. The design of a company site can be a long, expensive process. When efforts are focused in areas that are not effective, the entire endeavor could be a waste.

Even if you’re happy with your site, the need for testing still exists; there could be a method that brings more success than what you’re used to seeing. A successful business owner is always striving for more. In fact, a recent study by Monetate of their testing clients demonstrated that companies that are pleased with their sites’ conversion rates run 40% more tests than those who are unhappy, which may be the opposite of what’s expected. Furthermore, these companies test on an ongoing, or even daily, basis to ensure their site’s targets are met.

Areas to Test

There are many features and areas of a website’s design that can be tested. These include, but are not limited to:

  • Call to action buttons
  • Page layouts
  • Navigation structures
  • Content and copy
  • Special offers
  • Checkout process
  • Word choice (for example, using the word “free” in a title versus eliminating it)
  • Image selection
  • Security options, and more

A/B Testing

A/B tests involve testing two versions of a website with a single change against each other. The A version refers to the control and the B is the variation. By taking a look at live traffic, rankings, and conversions, A/B testing allows brands to measure a specific action or feature against another.

A/B testing is simple, straightforward, and easy to set up. The results are generally clear because of the lack of variation and the process is inexpensive. However, A/B testing has limitations. One of these pertains to the overall goal. If your company has a specific goal in mind that is clear and is accompanied by a specific user action, A/B testing is great. If however, you’re looking to measure the overall success of your site or are looking for the direction in which to head, A/B testing will not provide the results you’re searching for.

If your company has a specific end goal and wishes to test a single feature, an A/B test may be most appropriate. To get started, think about the feature you’d like to test and whether it is possible to change that alone to run a test. If so, reach out to a company that specializes in A/B testing or someone familiar with Web analytics. As a straightforward method it can be run without much difficulty.

Multivariate Testing

In other instances, multivariate testing may be a more viable option. A successful website requires a certain balance, a combination of design features, content, and navigation that lead visitors through a desired process yielding specific results – conversions.

Multivariate testing allows brands to test different balances of variances – including those listed above – while measuring traffic and drawing conclusions from the results. For companies looking to create a new site or to renovate an existing site, without a specific, single feature to test, multivariate testing is the method of choice. Its limitation comes in the form of a lack of specific results. While a test may clearly demonstrate that a certain combination of features works best, it doesn’t usually show which of the variables is most effective. It instead provides direction – the kinds of combinations that tend to work for your audience.

To create a multivariate test, choose a provider or online tool that provides the features you need. Google’s Experiments tool is a free option and allows users to test up to five separate versions of a page. Think about the features you’d like to test and what results you would consider to be “successful” (conversion rates and so on). From there, study the results and continue to change variables until you have reached a pinnacle. This will provide direction.

Testing is an essential part of a successful website, whether you’re a small company offering water therapy or a multi-million dollar retailer. Selecting the right form of testing is the first step. Think about your overall goals and look for ways to get started today. The time to test is now.



Case Study – Watters

Watters Logo | ReTargeter

Watters is the brainchild of designer Vatana Watters. For over 30 years, it has been the leader in offering luxurious designer bridal gowns, innovative bridesmaids dresses, classic special occasion dresses for mothers of the wedding, and adorable dresses for flower girls and junior bridesmaids around the world. Selling primarily at trunk shows and in third-party […]

Read more