The UBC Academic Search Google Scholar Blog, brought my attention to the article in June 10th BMJ, Search engines increase online journal use more than open access (not free, subscription required), BMJ 2006;332:1353.

While open access publishing does increase readership and usage, Researchers at the Center for Publishing at the University of College London, say that online search engines increase the usage more.

The researchers followed one specific journal, Nucleic Acids Research, between January 2003 and June 2005. In June 2003, Oxford University Press (publisher of the journal) allowed search engines and robots (Google, Yahoo, etc.) to access their website. The journal saw a 143% rise in usage from 2003 to 2005. According to the publishers the introduction of the open access publishing (in January 2005) only showed a 19% increase in usage.

Ok I have no doubt that opening your site up to search engines and robots will increase your site's usage, I do doubt the the actual percentage of increase.

Oxford opened up the journal's site to search engines in June 2003 and the difference in usage from Jan 2003 to May 2003 was compared to June 2003 through June 2005. According to the researchers open access publishing did not begin with Nucleic Acids Research until January 2005. However, they stopped monitoring usage after June 2005. Now I am not a statistics guru but wouldn't you think that one should compare data the first six months of the search engine data to only the first six months worth of open access data? Compare like to like. Comparing almost two years worth of data to 6 months worth of data just doesn't seem right.

This is a very brief article, and there is not much information on their methods for tracking the usage other than that they used a "deep log analysis" which collected "digital fignerprints of users specific internet sites" to track usage. It would be interesting to read more about their methods to learn more about importance of these statistics.