louisrosenfeld.com logotype

Home > Bloug Archive

Nov 22, 2005: Search Analytics or Search Log Analysis?

As I've blogged here before, Rich Wiggins and I are writing a book on search log analysis. Or are we? We're wondering if it would be better to use the term "search analytics" instead of "search log analysis," as commercial analytics vendors seem to be moving away from relying on logs.

For example, Measure Map and Google Analytics, if I understand them correctly, don't mess with logs at all. Instead, you insert a few lines of code in your page templates which enable these services to monitor the stream of users' actions--including searches--continuously. So, no logs, though you might still gather the same kind of data that search logs would produce (and then some).

Another reason for switching rather than fighting: the term "search log analysis" is a familiar one in the IA world. But where else? If you Google it, you'll find 516 results. "Search analytics" retrieves 19,500 results.

Is there a strong reason (or any at all) not to switch to "search analytics"?

email this entry

Comment: Prentiss Riddle (Nov 22, 2005)

I've been grappling with this myself and still haven't figured out all the connotations of the term "analytics". Does it *exclude* log-based analysis in most people's minds?

I'm increasingly rubbing elbows with people who would never think of analyzing logs (how 1995!) even though there are many things, such as RSS feeds, graphics, audio and video, which are invisible to the newfangled analytics tools.

Surely there must be systems which let you crunch data gathered both ways.

Comment: Lou (Nov 22, 2005)

Surely there *should* be, but I've not encountered any yet.

In fact, it's amazing how few search engine and analytics vendors' reps even know that search logs exist, much less could be analyzed. Perhaps I'm talking to the wrong people, but I'm getting the sense that the industry is much less far along than I'd hoped and expected...

Comment: Chris Grant (Nov 22, 2005)

I think I can answer some of those questions.

1. Traditional server log site behavior data has have been joined by web beacon data which is sometimes a log, sometimes a database. So yes, I would abandon the "log" term. But I think you need something to signify that this is basically artifact data.

2. It's possible to do continuous, real time analysis via logs. That feature is not exclusive to the web beacon or page tag method of data collection; in fact many of the beacon services don't offer immediate reporting.

3. Logs are not obsolete for web traffic analysis although some people would have you think so. There are advantages to logs, as you say. One big one is the history contained in old logs - instant baseline and some pre-post studies too.

4. How many people searching for "search analysis" are looking for what you're writing about? The term "search log analysis" may be reaching as many of your intended audience members as the less specific term does.

5. "Web analytics" does not exclude logs in most people's minds, I would say. There are people who strongly feel otherwise but in my experience they tend to be salespeople working for the web-beacon analytics vendors, or people who've never really tried to see what can be done with logs.

6. Yes there are systems that let you crunch data gathered both ways, but not many. The big newcomer vendors use beacons exclusively (they make their money hosting the service). The big old established ones tend to offer both. WebTrends does and it even allows both types of feeds to go into one report.

7. The industry IS much less far along than you'd hoped or expected. I'm daily amazed by this myself. To find advanced web traffic analysis (or what I consider to be advanced) you have to go outside the so-called industry to vendors like SPSS.

Comment: Andrea Wiggins (Nov 23, 2005)

Terminology is all. As I understand it from the (fast-growing) web analytics industry talk, the term "search log analysis" means that you are using on-site search logs (it needs a sexier term, no doubt) to gain insight to users. Exactly the thing that you spoke about at the SI Info Architecture class last week.

On the other hand, "search analytics" means that you are using search engine referral information (referring engine and terms used) to segment users, and then examining those user segments for trends in behaviors that can be attributed to referral source in some way.

"Search analysis" comprises both.

I would suggest using a variant on "search log analysis" that uses something nicer than "search log" though I haven't any brilliant suggestions at the moment.

Comment: Mike Steckel (Nov 23, 2005)

I strongly prefer "search analytics." I think this is much better positioning for the book (and its authors) long term. Log analysis might describe the current methods, but I think in the future more tools will be available for use. With SLA, you tie the label too closely with the tools available today and limit extendibility.

I also assume you will be talking about more than just search logs, but discussing some emerging trends beyond log analysis mentioned above also. "Analytics" will appeal to a much broader audience.

Comment: Nick Finck (Nov 23, 2005)

Web Analytics is the term I have always used that encapsulates things such as log analysis, search analysis, and sometimes even said to include user analysis. I don't think we need to reinvent the wheel here, there is a whole profession for this type of work:


Web Trends has been doing it for years, though they typically rely on log file analysis as a base.. still a lot of their products use embedded JavaScript just like Google Analytics does. There are several others in the game as well from as small as Mint to as big as WebSideStory and Omniture.

Comment: Jens Meiert (Nov 23, 2005)

Web Analytics is a more generic term, but it seems absolutely appropriate. Search Analytics is too specific, while Search Log Analysis is somewhat unspecific, and not that generic like Web Analytics.

Comment: Edward Vielmetti (Nov 23, 2005)

Remember that the analytics you care about also encompass referrer data, which makes up at least half of the relevant traffic I get on my weblog.

(Lou, some day I will show you measuremap up close because I now have some real data)

Comment: Tony Byrne (Nov 23, 2005)

I would say "search analytics" for sure. There is a risk of possible confusion with SEO people, but it's simpler and as others have pointed out more inclusive than search log analysis (which also sounds vaguely scatalogical to me...but that's another issue).

Comment: Lars Marius Garshol (Nov 25, 2005)

I think the benefit of "search analytics" over "search log analysis" is that the former says what it is, whereas the second focuses more on a specific approach. As you write, the server log isn't necessarily the only (or even the best) source of information for analysing search behaviour. So why name the field or problem area after an artifact of the web server?

Comment: Lou (Dec 1, 2005)

Thanks all for your excellent advice!

It's looking like we'll go with "search analytics". Seems like a fairly strong majority favors it...

Add a Comment:



URL (optional, but must include http://)

Required: Name, email, and comment.
Want to mention a linked URL? Include http:// before the address.
Want to include bold or italics? Sorry; just use *asterisks* instead.

DAYENU ); } else { // so comments are closed on this entry... print(<<< I_SAID_DAYENU
Comments are now closed for this entry.

Comment spam has forced me to close comment functionality for older entries. However, if you have something vital to add concerning this entry (or its associated comments), please email your sage insights to me (lou [at] louisrosenfeld dot com). I'll make sure your comments are added to the conversation. Sorry for the inconvenience.