Why do search analytics continue to be so rarely used?
We're wondering: at a time when both user experience design and web analytics are taking off, why do so few people analyze their local site's searches? This doesn't really make sense, does it? Or does it?
Have a theory? Please share your thoughts here, and maybe we can get to the bottom of it:
Comments
My take:
1. Lack of familiarity with the value of search analytics
2. Lack of tools and task knowledge (the availability of your script notwithstanding)
3. No pressure (internal or external) to apply standard quality-control metrics to the search experience
4. Perhaps some reluctance to be user-driven... after all, facts can provide inconvenient truths (to paraphrase a famous American).
Posted by: K.G. Schneider | June 19, 2006 03:52 PM
We ask all of our clients whether they log and analyse search activity, and point out the benefits to them. The most common response is that they don't do it, have never thought of it, and are pleasantly surprised when we explain the benefits to them. In short, we get that "That's clever, why didn't I think of that?" response. So my take is that it simply never occurs to people to do this.
Posted by: John Wood | June 20, 2006 04:40 AM
Looks to be a very interesting book! I agree with John Wood that often there's a "why didn't I think of that" factor - the idea had not occurred to clients. Perhaps this is also because the site search is something that's handled by the IT group and not really considered part of Marketing. (Ditto for the File Not Found page.)
Another possibility is that marketers/designers don't see how the information can be acted upon. But in truth, the hard questions are some of the most useful ones: like, "why is everyone searching for our old product names..." might compel the client to add a note to the home page that redirects folks to the equivalent new product. It's actually a very creative and fun process - but marketers/designers might think it's all about number crunching!
I recently co-authored a book ("Search Engine Optimization: An Hour a Day") in which we touch upon the subject of utilizing internal search data for SEO to get a better understanding of keywords and the audience population. I look forward to learning more on the subject!
Thank you!
Gradiva Couzin
http://www.yourseoplan.com/
(sorry - no search on our site!)
Posted by: Gradiva Couzin | June 20, 2006 11:01 PM
To add on to KG's great list
5. Perception that search is a technology problem
6. It's more fun and exciting to focus on possible solutions than to spend time fully analyzing the problem
7. Belief that focus should be on how to get people to enter the types of queries we want them to enter, not tweaking the site to better support the types of queries they actually enter
Posted by: Jeff Lash | June 23, 2006 08:52 AM
Another reason might be a legacy of how smaller web sites work: web traffic loggging is built into all web servers, and most ISPs provide some sort of simple tool for analyzing the traffic -- but search engines are a separate application from the web server, to be installed and managed as part of the site's functionality, not the server's. Oftentimes the search tool is not mission-critical to the site's success, and is perceived as a nice extra feature. Many web sites don't have search engines at all. Many web sites don't even need search engines. So for many webmasters and web developers, web analytics are normal but search analytics aren't even applicable.
But for those web sites for which search is integral to the site's success, can you really assume that it is true that search analytics are overlooked to a greater degree than web analytics? In my experience, clients who don't analyze their search terms don't analyze their web logs very much, either.
Posted by: Christopher Fahey | June 26, 2006 11:57 AM
Simple. Most companies leave web analytics to the technical people (the "Webmaster" for example). THese technical people have no clue what the reams of data mean. So if you ask them for analytics, they say "we got 10,000 hits last month" or something like that. It's like if someone asled me to do financial analysis on reams of numbers. I'd say, "well we had a million dollars in sales last quarter" and that's about it 'cause I'm not a financial guy.
There's a goldmine in analysis waiting in web stats. The solution is for marketing people to take ownership of web statistics and analysis. We marketers need to do the work and not rely on our techies for help. THis is too important an area to delegate.
Posted by: David Meerman Scott | June 26, 2006 03:37 PM
In my organisation, the main reasons have been: a paucity of effective analysis tools; organisational barriers to accessing log files; a project-focused approach to web development ("It's broken! Quick, fix it! ... [18 months later] It's broken again! Quick, fix it!") rather than incremental development and refinement.
That said, we and other Australian universities are starting to make some headway in the field of search analytics -- see, for example, the several search-related papers and posters being presented at Ausweb06, the national WWW conference in July.
Posted by: Margaret L Ruwoldt | June 28, 2006 02:12 AM
I feel that yes, webmasters and technical people that are close to the data have failed to make the sale on why numbers are important, and how they can help drive the success of a site.
It is the responsibility of the numbers person (or webmaster) to show the data, discuss trends, voice the trends, and then fold the data back into the site in a way that improves content. For example, once you know the top 10 searches why not feature them as live links on the site homepage?
These direct connections between data and features, to the point where the data becomes the feature, can have a dramatic impact on the perceieved value of search/web analytics. Site stakholders can see the self-managing properties of data-driven content features; and users spend less time searching or typing becuse your using their imput (or tracked behaviors) to improve their experience.
Posted by: San Juan | July 5, 2006 01:53 AM
Why hasn't search analytics taken off?
Maybe:
1, Don't have the tools to generate the search logs. Also (as indicated previously) search logs remain in the hands of technical people, and don't always filter through to IA's, and marketing types.
2, Search log analysis is a pretty time-consuming/ manual plod, and probably isn't that much of a popular activity (for some).
3, Ambiguity & context - search logs as a raw list on a page are so slippery and ambiguous, that trying to second guess people's original search intention is very tricky. However this is also why search log analysis is such an interesting topic in the first place.
4, Perhaps there's a certain anxiety in what might be discovered that tests the assumption of what users want from a site. Imagine a site with in-built expectations for users, only for them to be undermined by search patterns that reveal a different emphasis.
Posted by: Jonathan Snellgrove | July 18, 2006 12:49 PM