You Cant Fix What You Dont Think Is Broken

As powerful and important as web analytics is, web analysis can often hinder rather than help a web site or business improve. While analytics data is fine for telling you the "what" but not the "why" -- so beware of using metrics to make assumptions about your customer behavior, preferences or your site's performance.

Consider this situation. Retailer sees that the site search box is rarely - if ever - used. Retailer concludes "Our customer doesn't use search. We don't need to worry about site search optimization because nobody uses it anyway." Now efforts to improve searchandizing (the way you merchandize products in search results), personalization based on search behavior and improving site search usability become low priority or are disregarded altogether.

Suppose the "why" is the site design has camouflaged the search box. It's just not where folks expect it, or it's too subtle:

Examples:

In this case, a retailer could tweak the design, and revisit site search stats 1 one to 3 months' time to see if there's a difference. This doesn't have to be split-tested -- you could simply measure the before and after.

Of course, it may be a good idea to test different search box sizes, designs and placements using a website optimization tool like Google Website Optimizer, Omniture Test and Target or Sitespect.

The search box scenario is only one example. This could easily apply to any design element/feature of your site that you conclude does not get used because of lack of customer interest. If any of the following's use is below what you would expect for your site, investigate "why" and ask yourself if design or usability is to blame:

• Search box
• Navigation menus (top, sidebar, footer, filtered navigation, visual navigation/AJAX menus etc)
• Contact us forms (link hidden? form design problems?)
• Live chat (interrupt too soon? link hidden?)
• FAQ (customer service constantly answering questions already addressed on your website?
• Wishlist (do you require registration?)
• Cross-sell/upsell (too much choice? not enough choice? irrelevant suggestions?)
• Gift finder (buried in site? difficult to use?)
• Email/newsletter signup (do you show a privacy statement? do you ask for too much information? is the link buried?)
• Customer review participation (is the process difficult? what are the incentives?)
• Customer feedback (is the link easy to spot? are your surveys too long?)

It helps to start with an expectation/goal/industry benchmark of what you expect usage of various features on your site to be. Then look for the "what" in your analytics. If they're way off from what you expect, think about the adjustments/tests/surveys/fixes you need to improve them.