Diving Into Site Search Analytics: Interview with Louis Rosenfeld

Aug 11 2011 by Jacob Gube | 7 Comments

Diving Into Site Search Analytics: Interview with Louis Rosenfeld

Data from internal site search queries performed by your users is a gold mine of information. This information can be used to guide site improvements related to design, information architecture, content, and more.

Site search analytics (SSA) is an emerging form of site analytics that studies search query data to discover patterns in site use.

In this interview, we discuss site search analytics with information architect consultant Louis Rosenfeld, a groundbreaking persona in the field of SSA, whose list of accolades include the authoring of Information Architecture for the World Wide Web published by O’Reilly and, most recently, Search Analytics for Your Site under Rosenfeld Media.

What is site search analytics in a nutshell?

Louis Rosenfeld (LR): Your users are telling you what they want from your site — in their own words — when they use your site’s search engine. Are you giving them what they want? SSA is simply a set of tools and methods to help you harvest users’ queries, test and measure how well they’re performing, and actually see how well you’re serving users. That’s it in a nutshell.

How is site search analytics different from SEO?

LR: In SEO, you’re looking to learn from the web queries that direct — or ought to direct — users to your site. In site search analytics, you study what users search once they reach your site. At that point, their queries are far more qualified and relevant to your organization and your content, so you’ll learn a lot about how to serve users better once they’re at your site.

What are some things a site owner can learn from analyzing search queries on her website?

LR: Here’s just one example: she can learn that her site is completely failing to provide users with critical content — even if the content is already there on the site.

She’d do that by identifying and analyzing the common search queries that retrieve zero results, then determining if the content was indeed there or not. If it was, then she’d be in a position to see if the problem was with the search engine’s configuration, or with the content being jargon-y, poorly written, or poorly tagged.

As you can see, SSA is a good diagnostic tool. The site owner could then fix the problems or, in a large organization, wield the data as a highly persuasive tool for cajoling unruly IT staff and content owners to do the Right Thing and fix whatever problems she’d identified.

There are many other forms of diagnostics covered in the book, all of which can help you fix and improve your site’s content and navigation, as well as search performance. Additionally, site owners can use SSA to develop search-related metrics. Those come in especially handy when you want to benchmark, monitor, and optimize your site’s findability.

How can site search analytics drive design decisions?

LR: SSA can help site owners determine which content should be given the most prominence. For example, in the book I show how analyzing query data can expose the seasonal nature of users’ information needs. When you know which content matters when, you can feature content on your main page to match those seasonal needs.

Another example: most sites suck at supporting contextual navigation within and across their deep content. If you analyze the queries that originate from deep content, you’ll get a sense of the "desire lines" you should support — in other words, the navigational paths users wish were there.

What are some counter-intuitive results and patterns you’ve derived from studying search queries?

LR: Most of what we learn isn’t so much counter-intuitive. Like all forms of user research, it’s simply stuff that wouldn’t have occurred to us if we hadn’t looked at the data.

For example, the Financial Times noticed that many queries included dates. Assuming that users wanted articles that were published on or near particular dates, FT simply made sure that chronological sorting and filtering of their search results worked well.

FT also noticed that many queries were for the names of people and companies. So they configured their system to allow users to filter by names as well.

These are simple features that provide a lot of value. But, like a lot of simple, good things, sometimes you don’t think of them unless you explore and analyze data. Or, if you’ve got a boss or colleagues to convince, sometimes you need evidence to support you. SSA is hard — and very convincing — data.

What tools can be used to gather data for search analytics?

LR: This doesn’t have to be a complex or expensive proposition. A free tool like Google Analytics can do some basic SSA reporting (most people don’t know this; it takes a little bit of configuration to get it to work with your site’s search engine). More importantly, you can then grab the data and drop it into a spreadsheet for more customized analysis (here’s an Excel spreadsheet you can use for analyzing queries).

Why should a site owner invest time/resources in site search analytics?

LR: What’s important to remember is that even an hour per week doing even the most basic analysis will teach you something of value. Here’s why: let’s say you only have time to analyze and test, say, your top 25 queries. Well, if you map queries out from most frequent to least frequent, they look like this:

Source: SSA009: Figure 2.4.

This is called the Zipf Distribution — and it’s pretty much true for everyone’s site: a few queries handle a large portion of the traffic. The rest — the "long tail" — are fairly esoteric.

Here’s another way to look at this data:

Source: SSA010: Table 2.1.

This table shows how, in this particular example, the most frequent 14 (of tens of thousands of) unique queries accounts for 10% of the site’s search traffic. To get to 20%, you only need to analyze 42 queries. To get to 30%, only 98.

So your efforts are quite scalable, and a little analysis can go a long, long way.

If I wanted to start analyzing search queries, what’s the best way to get going?

LR: Cheaply. Set up Google Analytics (free) on your site. Teach it how to parse out your search engine’s queries. Use GA’s standard reports, but be sure to download the data into your favorite spreadsheet application and… play!

More on Louis Rosenfeld

If you’re interested in learning more about site search analytics, get your copy of Louis Rosenfeld’s new book, Search Analytics for Your Site.

Also, check out Rosenfeld Media’s other books, such as Luke Wroblewski’s Web Form Design. Rosenfeld Media will be releasing 13 new titles by 2012, so make sure to stay tuned!

Louis Rosenfeld’s blog is at www.louisrosenfeld.com. Follow him on Twitter @ louisrosenfeld (and @RosenfeldMedia).

Exclusive Discount

Get a 15% discount on your purchase of Search Analytics for Your Site by using the following discount code:

6REV

Related Content

About the Author

Jacob Gube is the Founder and Chief Editor of Six Revisions. He’s also a web developer/designer who specializes in front-end development (JavaScript, HTML, CSS) and also a book author. If you’d like to connect with him, head on over to the contact page and follow him on Twitter: @sixrevisions.

7 Comments

Robin Jennings

August 11th, 2011

Site search analytics is so vital when you have a large cumbersome website with archives upon archives and wide ranging topics.

I don’t think I’ve been to a news website where I haven’t had trouble finding information before.

Louis, I don’t think you will be ever out of work!

Henry Louis

August 12th, 2011

I have known about the importance of site search optmization but didnt know a book can be written on it, Congratulation Louis. This article had already shared pretty insightful tips – interestingly google analytics can be used for SSA. I would even agree with Robin that this strategy is much more important for websites which have hell lot of archives.

Christian Okpada

August 12th, 2011

Great awakening. Very useful

Lou Rosenfeld

August 12th, 2011

Thanks for the kind words! Yep, many Google Analytics users don’t know that they can do this. It’s a little tricky to set up, as you have to configure GA to understand how your particular search engine presents queries; here’s how they say to do it: http://www.google.com/support/analytics/bin/answer.py?answer=75817

Tim Leighton-Boyce

August 12th, 2011

I don’t think the strategy just applies to web sites with huge archives. Any site with a decent number of visitors will discover that the diversity of people’s expectations about navigation and language is a very big factor too.

The great thing about site search analysis is that you get to see the actual language that your visitors use. Studying anything which allows you to listen to your visitors using their own words is extremely valuable. I find it’s often very revealing.

I’m looking forward to reading a whole book devoted to the subject.

Irina

August 15th, 2011

SSA is definitely a great way to gather important information about the visitors, their intents and needs. Didn’t know GA can be used for SSA, thanks for the useful tips.

Matthew Wehrly

August 22nd, 2011

Hi Jacob,
Thanks for great read! Off to grab a copy!

Matthew Wehrly

Leave a Comment

Subscribe to the comments on this article.