A General Critique of Google Insights for Search Forecast

A couple of days ago, I had a conversation with a day trader friend of mine about investing on stocks when good news come out about a company. He told me that when the news are out, it’s because insiders have already made their money. As a result, no real gain is going to happen after, which means that it is pretty much time to sell!

I couldn’t stop from drawing parallels with the search engine market, especially the buzz around Google Insights for search forecast. Professional search engine marketers know that search engines are strong pull technique, i.e. people search for things they have already heard of, i.e. chances are that there are already stuff on that subject and that some of those stuff are well ranked on search engines.

This is precisely where lies the dilemma: on one hand, marketer want to look for important keywords to use as promotion channels but on the other hand, these keywords are most probably already taken by SEO experts who are not going to let their position go away. In such a scenario, it is only natural that people turn to crystal balls, in this case Google Insights.

Less than two months ago, I searched for ‘Microsoft‘ on Google Insights and the crystal ball told me that interest for that keyword was going down. Well, I asked the question again and the crystal ball has changed its mind! Take a look at the results then:

Then: Microsoft was dead according to Google Insights

And take a look at the results now:

Now: Microsoft might resuscitate according to Google Insights

Well, it all looks like we live in a world in constant change, which means that  doing extrapolation of past results to predict the future will mostly give wrong results. So what is the point to build such a tool, especially if Google is to have that warning below their forecast chart that says: “Forecast values are based exclusively on the extrapolation of past values”? Marketing!