Updated: Fixed URL to Insights.
One of my colleagues in marketing pointed to the new Google Insights service that helps you analyze the search terms people are using at Google. I've heard people talk about doing this, but it's involved digging through the raw search logs.
We're digging around with Insights at work to check our products against the competition, so this is a nice opportunity to give a quick overview.
Google Insights provides a simple UI and some basic graphics to help track the terms. The basic elements are an interest-over-time graph that can be adjusted for various time spans. Then there is a regional interest list that shows relative popularity of the term around the world with a highlighted world map - this can also be drilled down. Finally there is a related searches and rising searches analysis for terms related to your terms.
After the basics, you can do some fun things like comparisons of multiple terms to see the relative popularity. Or you can shift the focus from the search term to time range (and compare the single term on multiple years, for example). Or you can focus on a region / multiple regions to see how a term compares.
Oh, and the term can be something you type, or something from their list of categories. Or you can look for your term within a specific category.
The Google Insights team have built several help pages that describe these analyses and results. And there are a number of entertaining examples on the Insights home page that give you a flavor for the complexity of searches.
But how does it look? How does it work?
Of course, one of the first things the digiterati do in a new search tool is the ego search. Sadly, there is not enough search volume on "jack vinson" to show any results.
If I look for something in which I have interest, there is something to show. A default check for knowledge management shows a normalized interest over time graph. It's been fairly stable over the last year, but if I go back as far as I can (2004-Present), the interest in KM appears to be steadily declining, as shown below. The letters on the graph indicate news stories that Google has connected with the term. In this case, these are heavily slanted to technology stories / press releases.
What about a comparison of a term like knowledge management with web 2.0 since 2004? This may not be fair as web 2.0 has become immensely popular, but it's interesting to see that it has only overtaken KM since 2006. (KM in blue; Web 2.0 in red)
I am not impressed with the regional interest information. Sri Lanka and Bangladesh are the highest relative hits for some of these phrases with the United States with a distant volume. Maybe the search term needs to be something massively popular for this regional data to make sense. Clicking through to a country gives you regional breakdowns within that country, if they are available (states, provinces, etc). It also adjusts the interest-over-time and the related searches information to include only that country or region. I'd like to see the ability to focus on several countries at once, such as North America or Europe or the European Union.
The related terms for knowledge management make sense, although the top related term is project management, which doesn't. Reading the explanation for this part of the analysis, it is looking at search terms used before and after searches for your terms. So, people looking for KM are also looking for information on project management.
To wrap up, let's look at one of their canned searches. This one shows the relative popularity of "tour de france" over the last five years. Blue = 2008; Red = 2007; Yellow = 2006; Green = 2005; Purple = 2004. The high peak in this graph is 2004 and 2005 - the last two Lance Armstrong years.
Questions / Comments
What's the deal with the regional information? I can't believe some of the regional splits, but then others seem like they aren't so far off. I am guessing some of this has to do with the volume of searches.
It appears that many things I search show a general decline from 2004 to 2008. Our product names, competitor's names. Is this an artifact of something else in the data?
You can search multiple, similar terms with the plus sign (+) as an "or" statement.
Search terms appear to be full text. No need for quotes.