Amadesa

As the first in a series of entries, Amadesa is going to provide you with industry best practices and helpful hints! However, the first step is understanding what to do before launching an optimization. We will break this topic into several sections as it is one of the fundamental (and most important) steps in testing.

Before you launch an optimization you want to be prepared. Things to keep in mind:

• Analyze your web analytics data and identify key pages or activities that you can test to improve overall business objectives
• Determine the appropriate number of test versions based on your traffic volume

There are other points to consider, but with these initial data points in mind, you’ll be on your way to being prepared to launch your tests. Let’s look at these in more detail.

Analyze Your Data

Analyze your web analytics data to identify highly trafficked pages with high bounce rates and/or high exit rates. Your web analytics data is one of the keys to uncovering testing opportunities. Prime pages to test are those that have the potential to drive revenue, enhance the user experience and are being abandoned by your visitors; always keep in mind the overall business objective and strive to continually improve your Web site.

Highly trafficked pages with high bounce rates are a good place to start in identifying where to test. These are pages where visitors are not staying on your site; they are getting to the page and immediately leaving. This is a great place to start testing to see how you can make your page stickier.

For example, one page that typically has a high bounce rate is your home page. Users may abandon because they did not find what they were looking for – this is where you need to help solve their problem. Users should be able to land on your home page and be able to identify with something on the page or know what to do. Have you ever arrived at a home page and felt it was overwhelming or didn’t have what you were looking for? I would abandon, wouldn’t you? Instead, why not help users better identify themselves and relate to what they see on your home page. Think about how you can be more relevant to every visitor by testing alternative content and messaging.

Pages with high exit rates are also good places to optimize. Look at your path reports and see how people get to those pages; perhaps there is a trend that might tip you off as to why people are exiting. Looking at the path reports may also give you insight into opportunities to optimize higher up the funnel. For instance, if your cart page has a high exit rate, what can you test to improve click through to the checkout process? Frequently we see that value propositions, security reinforcement, clear messaging and few points of abandonment (links to pages other than the next page in the funnel) greatly affect cart page click through rate. Additionally, messaging on product detail pages may influence behavior downstream as users are ready to make their purchase.

Pages that contribute to the conversion path are obvious stomping ground for testing. Conversion contributors are pages in your funnel that facilitate or impede your visitors’ ability to convert. A product page is a perfect example of a conversion contributor. How the page is designed, what are the calls to action, etc. all impact the visitor’s next step. Testing something as simple as the “Add to Cart” button or enhancing promotions can dramatically contribute to a visitor’s desire to convert.

How Many Tests do I Run?

Determine the appropriate number of test versions based on your traffic volume. Traffic volume is important to consider when testing – the more traffic you have to your site or to a specific page you want to test, the more tests you can run. The key is you want to have enough traffic to reach statistical significance, so you can make an informed decision within a reasonable amount of time. If you need more than 3 months to achieve statistical significance, you may be missing out on other optimization opportunities.

For example, say you want to run an A/B/C/D test on your home page. Your site gets 5,000 visits, has a current conversion rate of 3% and you want to see a conversion rate lift of 10%. Based on this data you would have to run your test for 15 days to reach statistical significance at 95% confidence.

A good rule of thumb is you want to have at least 50 conversions per version per day. Amadesa has several tools to help determine the number of test versions they can run based on traffic volume including A/B and Multivariate Testing calculators.

Taking these initial steps – analyzing your analytics data and assessing your traffic volume – will help you prepare for a successful wave of testing. Good luck and get started!

In our upcoming posts look for topics such as other factors to consider when launching your tests include (specials, promotions, email campaigns, etc.), why is it important to track revenue even if you are not operating an ecommerce site and using your analytics provider to validate data. We will also detail how to analyze your tests once you are up and running.