nicolas leroy

Testing Google Trends for Websites on shopping engines

July 11, 2008

Since one month, Google Trends now shows traffic trends for websites, and therefore becomes a competitor of Alexa or Compete. Google is a bit secret about the way data are collected:

Trends for Websites combines information from a variety of sources, such as aggregated Google search data, aggregated opt-in anonymous Google Analytics data, opt-in consumer panel data, and other third-party market research. The data is aggregated over millions of users, powered by computer algorithms, and doesn’t contain personally identifiable information. Additionally, Google Trends for Websites only shows results for sites that receive a significant amount of traffic, and enforces minimum thresholds for inclusion in the tool.

And explains we have to be cautious when using the data:

Keep in mind that Trends for Websites is a Google Labs product and that we are experimenting with ways to improve the quality of the data. Because data is estimated and aggregated over a variety of sources, it may not match the other data sources you rely on for web traffic information.

Another limitation of this tool is that subdomains are currently not considered:

At this time, Trends for Websites will only show results for websites at the domain level. If you enter a subdomain such as subdomain.example.com into the search box, it will be converted to example.com instead.

That means it’s not possible to track data for Yahoo! Shopping US (shopping.yahoo.com) as data are aggregated on the yahoo.com domain. That also means that the traffic of Shopping.com not only reflects the US traffic but also the worldwide sites (fr.shopping.com, uk.shopping.com…). Annoying…

I compared the data for some shopping engines between Google Trends and Compete, and they can hardly be more different :) A recent article on TechCrunch explains how those systems are flawed and how Firefox (yes, the browser :) ) could bring a smart solution to this problem:

None of these services are particularly accurate (as can be seen by the fact that they almost always disagree with eachother). The problem is simply gathering enough data from enough users to be able to draw a picture-perfect image of actual Internet usage. […] Ask Firefox’s 170 million (and growing) user base if they would like to opt in to anonymous data collection on their surfing habits. Then take that anonymized data and create very statistically relevant analytics reports for all websites. Only a small percentage of those 170 million users would have to agree to be tracked (Lilly [Mozilla CEO] said 1% is more than enough) to get useful data. There are Firefox users in every country, and the distribution is fairly attractive for worldwide analytics tracking. […] Of course, this would track only Firefox users, not IE, Safari, Opera and other browsers. And Firefox users as a group may have different surfing habits than the Internet as a whole. But as Firefox usage grows more mainstream, this will become less and less of a problem.

So Google Trends for Websites is a nice tool to play with, but doesn’t seem (for now) a better solution than other existing analytics tools. We can only hope Mozilla will do quick progress on its Stealth project.

Comparison of some french shopping engines between Google Trends:

and Compete:

Comparison of some US shopping engines between Google Trends:

and Compete:

(by the way, it seems that Google Checkout Trends, the shopping-related trend tool from Google, is currently not working ; I got 502 errors…)