The virtuous (or vicious) Google data machine
Ten years ago this September, Sergey Brin and Larry Page were about to set up offices in a garage. Their Web site, Google.com, delivered results for 10,000 searches a day. By 1999, that figure had jumped to 3 million. Today, Google processes hundreds of millions of searches each and every day. If this is a staggering number, consider that behind each and every search query is a set of consumer data, and it is access to this data that puts Google into a league of its own.
Firms that sit on a massive lot of data traditionally have two choices: a) use it for internal purposes or b) sell it. In the Google era, there is another option: c) give some of it away for free. Some might say that Google's forte is search; I would say that it is the art of giving services away in exchange for data. In a twist of irony, the free service known as Google Trends is actually a glimpse into the data itself.
Launched just over two years ago, the tool provided a clean snapshot of search patterns over time for any set of keywords. Key aspects of the tool include:
· Scaled search volume numbers. A year and a half since we accepted that our beloved Overture Tool would be no more, it appears that we will never again access actual search volume. Google Trends' own search volume index scales volume based on the first term entered, with 1.00 indicating the average search volume of that term for the selected time period. All other terms entered are then scaled.
· Data exports. One of the more recent updates is the ability to then download scaled search volume data to a CSV file with relative scaling relative or fixed scaling (usually January 2004). The download also delivers region, city and language data.
· Trends for Web sites. The latest development is the ability to see an estimate of traffic to websites, percentage of visitors from a particular region, sites users also visited and the keywords the users are likely to search for.
Last week's foray into Web site measurement set off a new set of bells. Journalists have started to write obits for firms whose primary mission is to sell this type of data on a subscription basis, such as Comscore, Nielson Net Ratings, Hitwise and Compete.
As Fred Wilson points out, this is not the likely cause of death for panel services. Most Internet marketers understand that short of looking at log file data, Web traffic services provide estimates based on a series of data gathering factors. This is why many firms triangulate data – that is, subscribe to multiple services. Since Google Trends for Web sites data is derived from Google itself, the patterns are representative of Google users, but not of the general Internet-using public. It then becomes just another source of data to add to the mix.
So why all the brouhaha over Google Trends? Because this public tool is just the tip of the data iceberg, which includes Ad Planner, a free research and media planning tool now available by invitation to advertisers who require access to a deeper layer of data, such as demographics and sites. And so the virtuous (or vicious) cycle continues.