Cool custom report: What on-site search terms led to an exit?
Suppose somebody on your site searched for something (using on-site search) and then left the site directly from the results page. Wouldn’t you want to know what the search term was?
This is a semi-easy custom report that will give you just that — a list of search terms for site search results that were followed by an exit.
To do this report, you have to know the URL of the search results page and the name of the parameter that contains the search term. If you set up your site according to the WebTrends book you’ll have WT.oss as the name for the search term parameter, but the parameter could also be “keywords” or “q” or anything else. (You’re not collecting the on-site search term? Give it some thought; it’s worth its weight in gold.)
You need to create:
1. A custom visit filter that includes only visits where the Exit Page was the search results page. Note that I said visit filter, not hit filter.
2. A custom dimension based on “Query Parameter,” naming the on-site search term parameter that your site uses or WT.oss if it’s collected by <meta>. But this dimension has a twist — the “When To Collect Data” setting must be “Last Occurrence in Visit.” This is key, otherwise you’ll get all the search terms in the visit when you want only the one that “caused” the visitor to leave the site. Handy anti-confusion tip: put “last occurrence” into the name of the dimension to prevent mixups with the ordinary version of this dimension.
That’s all. Put together the report as usual.
A really good (I would even say essential) follow-up analysis is, for each term, the ratio of “exit instances” to the total number of searches using the term. The latter is of course available from another custom report, set up without a filter and an “Collect Data On All Hits” version of the dimension.
Related Post: What on-site search term led to this page?
6 comments
Hi,
Trying to set this report up but am confused with which parameter to use for the custom dimension:
[value to be based on] should equal ?
All the search options (phrase/keyword) don’t then allow me to chage the ‘when to collect data’setting?
Thanks,
Julian
“Value to be based on” should be “Query Parameter”. Then enter the name of the query parameter that your on-site search uses to display the search term. Could be “keywords” or “WT.oss” or “q” or whatever.
With “Query Parameter” as the basis for the dimension, the “when to collect” screen allows choices.
Sorry for the confusion. I’ll go back and clarify the text.
Thats great, is now working!
I’m getting confused as to which measure is best to put against [site search term] would you suggest that visits is correct?
Visits being the number of times a search terms is used whereas pageviews against a search term is more related to number of pages viewed after a search visit.
Julian, Yes, I’d agree!
So have you found anything interesting or helpful in the report?
I’ve setup three of your suggested internal site search reports and am expecting them to be very useful! From providing business actionable insight for our site UE team to improving our backend search results and indexing.
Cheers for your blog, been a great start into getting useful results out of Webtrends, as a previous Google Analytics user the change has been a bit painful:)
Good!
A story: using on-site search analysis, we found a major blunder by the merchandising people — they thought everybody used the industry term “swimwear” and that’s the only word they used. They were one of the few sites that sold a wide selection of swimwear year ’round so it was important to the winter vacation population. The on-site search showed additional terms that actually got no search results — “swim suits” and a couple other regional terms. They corrected the on-site thesaurus and also widened the vocabulary in the visible text and product names. In addition to getting the on-site searchers to the right part of the site, the copy changes influenced their external search rankings for those extra terms. So, business for these products went up two ways. And the people looking for swimwear also found tropical clothes in the middle of winter, so average sale size increased.
We love internal search.
Leave a Comment