Search Engine Numbers Are Immensely Deceptive

Quick, what's Google's market share of searches? There's no doubt that it's big -- really big. But just how big turns out to depend on how you define a search engine. Some numbers suggest that YouTube is the driving force behind the company's gains. But that has to leave you wondering what constitutes searching and whether the market share numbers are leading advertisers and the high tech industry to conclusions that may not be exactly justified.

Reported results on search engine use can vary so widely that you wonder if they're on the same map. Here's an October example from Net Applications:

Then there are numbers from comScore:

Quite a difference. One reason is that although many media outlets use comScore as a source, the company has been publishing its analysis of the U.S. search market, not the global one. But it also explicitly excludes from its core search numbers "searches for mapping, local directory, and user-generated video sites that are not on the core domain of the five search engines." It's another way of saying, for example, that Google's numbers don't include YouTube searches. Given that Net Applications doesn't mention YouTube, you might wonder if it is included under Google. Yes, Google owns YouTube, and yet something seems fundamentally different about looking for information, goods, and services, and searching for a given video. If you sell widgets, will advertising on a video site really make sense?

The difference between comScore's core and expanded numbers is significant. Look at these two tables of October data showing numbers of searches:

Not only are there big jumps, but you have to read the categories carefully. For example, Google Sites is the sum of Google and YouTube/All Others. Skip by that and you could end up significantly over-counting the company's already impressive dominance. As far as global numbers, it's impossible to accurate project comScore's numbers beyond the borders of the U.S. Furthermore, as is true with a number of the Internet measurement services, results sit on the habits of large panels of users. But people agreeing to be part of such a panel might well not be representative of consumer behavior on the whole, in which case the reported numbers could be complete hooey.