You would do better to measure Toolbar PageRank than to gauge market share on the basis of number of queries performed. There are many people searching for themselves and checking rankings at Google.

A large portion of that search activity flows through the ISP connections that Compete and other metrics companies sample for their market share estimates.

A much better gauge of search market share would be number of search referrals (broken down by organic vs. paid would be even better). Unfortunately, there is no way to get that kind of data if the search engines won’t divulge it.

Some large PPC providers publish their PPC referral data (in aggregate to protect client privacy) but they don’t fully disclose their methodologies and I suspect there are advertising budget biases in the data.

A better metric than queries-performed is number of visitors to search destination, although that data is not completely unambiguous. As long as you can get estimates for search subdomains, however, you at least have a much clearer idea of how many people use each search engine.

Nearly as many people use Microsoft’s search services as use Google’s. THAT is a number worth knowing.