I recently attended the eMetrics: Marketing Optimization Summit last week in Toronto. I sat in on a workshop on Web Analytics for Site Optimization given by Jim Novo from the Web Analytics Association.

At MediaTrust, we are constantly researching new ways to optimize websites and landing pages to improve conversion rates for our affiliates. This is a brief overview of some of the things I learned from Jim’s workshop:

What is Web Analytics?

Web Analytics is the measurement, collection, analysis and reporting of Internet data for the purposes of understanding and optimizing Web usage.

Build Sites with Web Analytics in Mind

Don’t waste time and money retro-fitting your website. Make sure you design your site with the latest technologies that support analytics. For example, if your website is a WordPress blog, upgrade to the current version so you can take advantage of widgets and plugins. There is a free widget for Google Analytics! Awesome! Personally, I try to avoid designing Flash sites because setting up analytics on them is too complicated. You can use flash elements in your site, but an entire Flash site is bad for web analytics.

Quality Traffic

Quality website visitors are more likely to be interested in your product or service and have a high chance of converting into an acquisition or sale. If you have a solid source for internet traffic, you will spend less time waiting for your analytics experiment to end. Quality traffic will convert more often and it will be easier to figure out what’s working and what’s not. Sites that have both quantity and quality traffic will collect data quickly and they get you to your next analytics milestone sooner.

How Much Traffic? A/B Split Test or Multi-Variate Experiment?

Not only do you need quality traffic, you also need a lot of it! Google says it needs at least 500 visitors per week for an A/B Split Test and at least 1000 visitors per week for a Multi-Variate Experiment. The reason for this is because, essentially, A/B Split Test is only two versions of the page. Multi-Variate Experiments make more versions from the combinations of the specific areas you defined. For example, if you mark off and test four areas, and each area has three variations… 4×3=12 pages. You will need more visitors for Multi-Variate Experiments to acquire enough data and really tell which combination is performing the best. If you take the Multivariate route, I recommend keeping the number of testing areas and variations low. The more areas, variations and combinations you have, the longer your experiment will take. As they say, “Time is money”, so keep your experiments quick.

Define Your Page or Site’s Goal

With landing page design, usually the goal is to get the user to convert. In affiliate marketing, conversion is when the user clicks and navigates themselves to the end of your process or basically does what you want them to do. For example, in a shopping cart site, you want the user to purchase the product. Once you define your site’s goal, you will probably realize that it’s too difficult for the user to use your site and reach the end of the process (the goal). Focus on the user’s experience. Try going through your site as if you are the user.

“Humans Track Information Like Animals Follow a Scent”

The users of your site are trying to get information or accomplish a goal. They are looking for a trail of links and buttons to click, so they can find what they want. Jim Novo calls this a scent trail. If the scent is strong, users will follow it until they arrive at your goal. If the trail is weak, they will either abandon your process (go back to the home page), or abandon the site (totally leave). Make the scent trail to your goals strong!