Google Trends has always been interesting and fun but it can be easy to question the results and their accuracy.
Google has taken steps to make the product better and more discerning. A post on the Inside Search blog from Google explains
You may have noticed it’s easy to get tripped up measuring the wrong thing using Google Trends. When you look up “rice,” are you measuring search interest in Rice University or the rice you eat? When you look up “Gwyneth Paltrow” how can you be sure you’re counting all the common misspellings? Starting today, you’ll find new topic reports to help you more easily measure search interest in the people, places and things you care about.
The example Google shows is interesting as they compare the terms Harvard and rice. As you might guess most people are not searching for rice the food if they are looking at comparing colleges. Rice University is a great school academically but it doesn’t get the same attention as Harvard. Rice the food however gets plenty of attention and an old result may look like this in Trends
What? More folks search for Rice University than Harvard? Not really. The post continues
Now when you start typing into the search box you’ll see new topic predictions. Type “rice” and you’ll see predictions for “Rice University (University)” and “Rice (Cereal).” This makes it easy to do a fairer comparison.
The resulting comparison looks like one might expect
There are many instances where you might imagine this capability will make Trends results much more useful and accurate. Misspellings and more will ultimately be covered in this offering.
Right now the service is in beta
Topic reports are a beta feature and we want to get your feedback. To start, you’ll find data for more than 700,000 unique topics from Barack Obama to football (soccer) to Hayao Miyazaki, and you can slice the data to measure search interest worldwide or in any of the following seven countries: Brazil, France, Germany, India, Italy, UK, and US. We’ll continue to improve the accuracy and comprehensiveness of the data, and as we do we’ll add topics and expand to support new regions.
What do you think?