From DM News' Special Report on Web Analytics: Determining Visitor Intent: Analyzing Search Terms Used Within Your Site
If your Web site offers an internal search feature, you might be surprised that it can reveal a wealth of information about your visitors' intents and expectations. Knowing what visitors are searching for is valuable insight toward Web content, site design and merchandising decisions. A good Web analytics tool with internal search capabilities is perfect for unlocking the secrets held within your site's internal search.
Now, don't confuse internal search analysis with search engine keyword analysis. Internal search relates to the keywords that people use while exploring your site-the searching they're doing once they're already on your site. Search engine keywords refer to the keywords that actually brought people to your site.
People Search for the Strangest Things
When you're analyzing your internal search queries for the first time, you'll probably be surprised by some of the phrases, terminology and common misspellings that visitors use in a search query-it can be quite an eye-opening experience.
While some word combinations are valuable in determining the language and phrases your visitors feel comfortable with, I recommend you start with an overview of internal search keyword volume before using segmentation for behavior analysis. In other words, get a clear grasp of the big picture view, and then drill down.
Using "Bobs Fruit Site" as an example, let's first learn the most and least popular search terms by analyzing two months of data ( For Figure A downoad this PDF file http://www.dmnews.com/cms/lib/6564.pdf by clicking on the link or cutting and pasting it into your Web browser).
An experienced analyst's observations:
1. Less than 1 percent of the visitors used internal search. It may be worth relocating the search mechanism to a different place on the Web site's pages and then seeing if search queries increase.
2. Why are visitors searching for balloons and t-shirts on a site that clearly sells fruit? Maybe they're looking for fruit gift baskets? The site owner ("Bob") might consider adding these sorts of things to his line of merchandise-after all, if enough people are searching for it, many of them might buy it.
3. The most popular search term is "FedEx". From this we might intuit that our visitors are having difficulty finding shipping information or that they prefer FedEx to other shipping companies. In this case, we will need to make shipping information more prominent on the pages that these visitors see most.
Are there more insights lurking in that internal search data? Certainly. But for now, we'll move ahead to confirm observation No. 3 by segmenting and following the behavior of the visitors who searched for FedEx.
Confirming Observations through Segmentation
Once we figure out which pages our FedEx searchers visited the most, we can decide on which pages we need to add shipping info.
Using a segmented view of pages with most visitors, we learn that FedEx seekers favored the "Recommended" page and "Organic Fruit" page. Combining segmentation and a path analysis, we know which pages are candidates for FedEx shipping information.
In Figure B, it seems clear that the Organic and Recommendation pages should offer shipping information. (For Figure B downoad this PDF file http://www.dmnews.com/cms/lib/6562.pdf by clicking on the link or cutting and pasting it into your Web browser).
Using Information You Already Collect
Internal search systems typically put the search keyword into a parameter on the URL (usually 'srch=', 'q=' or 'search='), which exposes the data for analysis through a Web analytics tool. Continued analysis can reveal whether the t-shirt seekers are new or returning visitors, how much time they spend on the site on average, which geographies they hail from, and a number of other criteria that can help us identify visitor intent and expectations. The analysis is valuable by itself or as a supplement to market research. The possibilities are truly endless by combining visitor segmentation with internal search parameter analysis.