By Brooks Bell
President, Brooks Bell Interactive
Testing has become a buzzword in online marketing in the last 2 or 3 years. Marketers have discovered that testing makes it possible to increase conversion rates by as much as 500% without increasing marketing spend! It’s an exciting trend, especially in the context of tightening budgets and declining click-through rates.
So, what is testing, and how is it done?
In theory, testing is simple: present your customers with at least two competing marketing experiences in your live campaigns, and then measure response (This is called a Control/Champion and a Challenger). Once you’ve determined a winner, you test again! By iterating your creative, you can incrementally improve on your original results by incredible amounts.
Traffic is the first T in Testing. According to statistics, your campaigns must be seen by hundreds or even thousands of customers to accurate detect which message is winning. This is called statistical significance. It’s the likelihood that you would have the same results 95% of the time if you repeat the test.
Not having enough traffic is the most common barrier to testing. Anyone can test, but if it takes several months to achieve statistical significance, then it may not be worth your time or investment.
There are several testing platforms including Adobe Test&Target, Google Website Optimizer, SiteSpect. They are powerful technologies, but should be incorporated once you’ve mastered the basics. These technology capabilities are minimum:
- You need to be able to split your traffic into each of your test groups. This means randomly exposing each customer to a test experience.
- You need to be able to track the behavior of your customers in each test separately using your web analytics tags.
Giving your testing program enough time is crucial. Testing takes at least twice as long as a standard campaign. There is more creative development, more planning and more analysis. Testing should not be treated like a single campaign or individual project. It’s actually a broader shift in thinking about marketing. ? In fact, the key drivers in your marketing results, such as your offer emails, subject lines, newsletters should all be tested constantly and even indefinitely.
When it comes to testing, you need to trust your data. This means that you need to have strong web analytics. It has to be so trustworthy that you are willing to take action based on the results, even if it goes against your intuition.
Lastly, testing requires the cross-functional skills of a team. First, a marketing strategist needs to decide what to test. Then, a designer and a copywriter need to create the campaign itself. Then, a web developer needs to build the page, tag it, and launch it. You’ll need the support of IT to split traffic to each of the test cells, and QA everything to make sure it doesn’t break anything else on the site. Lastly, you’ll need some help from a web analyst to pull reports and interpret the results. It can be a big effort, but it’s worth it!
Testing is a proven strategy that is still an untapped way to make marketing more measurable, accountable and effective. With lots of time, traffic, a strong team, trustworthy data, and good technology; you’ll have the right recipe for a winning testing formula.
About the Author
Brooks Bell will be presenting “The Power of Split Testing” at Conversion Conference East on October 4, 2010 in Washington DC. Her agency, Brooks Bell Interactive, designs creative solutions, online conversion strategy, and A/B test plans for clients including Chase Bank, Nickelodeon, ServiceMaster and AARP. http://www.brooksbell.com