I am fan of some of the information and insight I get from ZDNet. A November 24th, 2008 commentary suggested that leveraging search engine words to develop a text analytics practice would be helpful in determining “Questions and queries, such as ‘How did the analysts react to our Q2 earnings results’ or ‘Tell me about the Blackberry Bold’”. Truth be told, I was developing this back in 1998. And I know I’m wasn’t alone.
There are two types of on-site searchers: One that searches because they think it is easier to find what they want (almost always wrong), and the second who can’t navigate to what they want. Both can tell you a lot about the user’s intent, and over time, correlated with other web analytics and business intelligence unveil more opportunities than anything Cap’n Jack found on the Black Pearl.
Step One: Identify which users (or at least which percentage) of users fall into each category: The search first, or the navigate, fail and search. The user who navigates, fails and searches is literally telling you what they are looking for in THEIR OWN words. Not the words of your marketing department, nor words from PR – their own words. Knowing this means you can likely increase your conversion rates by adjusting your search to respond to the words your customers are using.
Step Two: If there is a seasonal bias, note it. For example, maybe in November-December users interact with your site differently. Drill down for smaller seasonality variances in search – and how they affect your business.
Step Three: Correlate your data with the rest of your analytics package findings and other business intelligence tools.
Now be careful. You need to be working with “correlations of confidence”, not just “correlations of convenience”. Put another way, just because the bars open when it gets dark does NOT mean darkness causes the bars to open. Most data can’t correlate enough to be used as a predictor. Then some can.
Take your time with this, and maintain it. It will be worth your while.
Okay. Maybe I missed a few valid points in the article, but I do agree with the summary: “This allows for enhanced analysis, understanding, and utilization of the information attained as well as superior reputation management”.