11 replies to this topic

[cre8pc]

Lou Rosenfeld and Rich Wiggins, in gathering information for a book on local site search analytics, invited 206 people to complete a survey.

Search Analytics survey results

Questions:

1. Where did you learn about analyzing queries produced by a site's search system?
2. We're surprised at how few people and organizations analyze their own site's search queries. If you agree, why do you think it's so uncommon?
3. Can you recommend any useful articles, book chapters, or books that are related to analyzing search queries?
4. Can you recommend any useful web sites or other web-based resources that are related to analyzing search queries?

The responses are interesting, esp. the verbatim comments.

I think that search optimization efforts would need to use this (SA) data and create a process in which the task is fully supported to help us understand how people think about information and navigation, but perhaps it may also help us understand a user's level of patience with results.

he SES Latino show in Miami last week showed how the Hispanic market was less likely to use such data.

Through my frustrating experience with search development, I became interested in IA and eventually, targeted content delivery. Now I think that the ideal state for enterprise search is to have content come to the user based on attributes such as role, organization, task, and other tags that tell us about the individual and what they are doing.

(To pick a few.)

Are you confused about search analytics? Do you study your data? If so, then what do you do with it? What tools do you use? Do you consider search data important to website development, and why?

[ukdaz]

Do you study your data? If so, then what do you do with it? What tools do you use? Do you consider search data important to website development, and why?

Kim

Just to clarify are you asking: do I analyse the results in my traffic web logs?

Daz

[Black_Knight]

Just to clarify are you asking: do I analyse the results in my traffic web logs?

Nope.

This is about the search box on your own site being a direct and natural source of customer generated enquiries and feedback. What did they have to search for (meaning it wasn't right there where they could see it)? What did they have to rephrase a search for because they still couldn't find it (indicating a disconnect between your search functionality and the user's ability to intuitively use it)?

[cre8pc]

What Ammon said

[rmccarley]

I just added a search feature to my site and it's from Google. Is there a way to see what people are typing? I haven't noticed.

And... uh... didn't really think about it until now.

So thanks!

[joedolson]

WordPress has a nice plug in which allows you to track what people search for in your WP blog: Search History plugin.

Provides a very useful means to keep track of internal searches.

Personally, I don't get enough site searches to bother using the data, but it's nice to have in case I do!

[ukdaz]

Thanks Ammon.

Have to admit I do not use a search facility on my own website. No excuse as I probably should do.

Daz

[Black_Knight]

When I was working in-house at Dooyoo.co.uk we made a serious commitment to mining the search data extensively several times. However, with a site that size, with so much diversity in content, it was such a massive job that it could not be just a regular part of the schedule. There were thousands of searches per day made, and trust me that the long tail was very visible.

What might have been nice would have been the time to spit them out into a database, then cross-reference the words in that database against the referral data (visits referred from external search) to find more synonyms and phrase variations semi-automatically.

It was very useful data, not least in profiling different types of customers, but always remember that search data is a partial dataset, only featuring data from those who will use search (a significant proportion of net users find search boxes difficult and unintuitive, never knowing the 'right words' to try). Search data will obviously be biased to show a lot more searches from the people who love search boxes than those who use it as a backup. It will also have more searches from those who need to 'have several goes' than those who are good at finding what they want in one well-phrased search.

[marianne]

Good Afternoon All,

Search analytics are enormously useful in enterprise space to determine how searchers make their way to your site. Prior to converting to MSN Search, Microsoft allowed limited querying against the SQL Server DB that served the www.microsoft.com search. In doing so, it was possible to look at what someone used, what results were delivered, and how they iterated their search based on these results. At the 2003 Information ARchitecture Summit, The Accidental Thesaurus presentation took this method a step further by using this iterative data to then revise the precision of the search engine through manipulation of the governing taxonomy.

To answer the original question of this thread, AOL found a big barrier to taking advantage to of search analytics with the privacy backlash to the release of partial data from their search engine. If it is not possible to disassociate any/all private data from the search log, then it should not be used for review. The other barrier is one that I would call "the embarrassment of riches" barrier. Disraeli called it when he is reported to have said: "There are lies, damn lies, and then statistics." With so much data about what people are using, reusing, rejecting, clicking on, and abandoning, a case can be made for just about anything. However, what I learned in school is that people don't know what they don't know and that impacts their search. They also change their search based on what they find along the way and may abandon a search because it is time for dinner rather than their frustration at not finding something or finding the wrong thing.

For me, search analytics are a small piece of a very large puzzle that I'm still working on.

[Jozian]

people don't know what they don't know and that impacts their search. They also change their search based on what they find along the way and may abandon a search because it is time for dinner rather than their frustration at not finding something or finding the wrong thing

Marianne - I agree with this.

Was discussing similar analysis people do on shopping cart abandonment last week. While the information is good, it is not the whole story. For instance, sometimes I just 'window shop' - fill up the cart with what I would like to buy, but know I'm probably not going to make the purchase... And sometimes I get ready to place an order, wonder if I have enough money in my account to cover it, and then abandon because I'm not sure.

The point you bring up is a very important one. We can track WHAT people do, but we can only infer WHY they are doing it.

Even when the attempt is made to ask the user why they took an action, the decision-making tree a user travels is rarely effective communicated, in part because we do not always know all of the factors we took in to consideration before taking an action. Some of it is subconscious

Still, just because we can only see part of the picture with analytics, doesn't mean we should close our eyes to what we can see.

-Jeff

[Jim Newsome]

Hi all,

Interesting posts. I would like to add some of our experiences monitoring internal search. We have used Google Analytics to monitor internal search for several clients (by passing the search terms to a user defined tracking variable) and this has thrown up some v interesting data. Black_Knight has a very good point about the length of the tail and the breadth of search data - the difficulty is, as it always is, analysing this data to meaningful, actionable results.

A key metric we use here is exit rates for search results pages crossed against the actual search terms. This has highlighted some major usability shortfalls in the search technology.

Jim

[bragadocchio]

Hi Jim,

Welcome to the forums.

Site search is an interesting place to look for information about the expectations of visitors and the types of things that they may expect to see on your pages, but can't find.

For sites that focus upon lead generation, via phone calls and email, it also doesn't hurt to pay attention to the questions that people ask when they contact you, to see if you can answer their questions better with your web site. Not only are the questions asked helpful, but also the language used in asking those questions.

In addition to site search, I like looking through log files to see the actual queries used to find the site. I've run across some interesting questions posed in queries, that were partially answered by information on the site, but which could have been answered more fully, and helped those searchers and others better. Being able to see those questions is helpful.

WordPress has a nice plug in which allows you to track what people search for in your WP blog: Search History plugin.

I've used that plugin, and it's been helpful in understanding what some searchers are looking for when they visit. It doesn't get a ton of entries, but I always find the ones that it does provide to be informative and useful.