People may just need one "right" answer, but the "right" answer for you may not be the answer I'm looking for, or the page I'm looking for may not exist. Since the engines cannot read between lines (or between words) to see exactly what the person is looking for from the search term, it's good to have multiple results.
Still, I think results on 2nd page and after are waste. If engines limit the number of search results to just top 10, it would save time and energy. ;)

Nice post! And thanks for all the nice words.

For what's it worth (and because WP trackbacks may not be working with MT), I've posted a detailed response here.

There was one point I didn't address in my response. I didn't talk about the way we collected our data.

It's possible there's a bias issue that is affecting our results, but I think it's a low probability. The supposition I came to was from a collection of studies of people doing real tasks on real sites.

None of the studies involved using top queries. Instead, they were all watching folks use the sites for things they needed to achieve on their own.

So, if there's a bias, it's unlikely it was introduced by the method by which we looked at the Search data. And since this is across multiple studies with a variety of sites (from e-commerce to medical information), I don't think it's a selection problem.

As for Andrei's taxonomy, I think that's great for dividing up queries, but I'm not sure it's a good way to think about goals. Queries are an artifact of the crude type-something-into-a-box-and-cross-your-fingers approach to effective searches.

In our work, queries don't really match goals very well. They are just a manifestation of the user's inability to clearly state what they are really trying to do.

Think of the average query as the equivalent of a three-year-old complaining of stomach cramps. They just don't have the skills to adequately describe what they need.

Keep up the excellent work...

Jared

Motoko, I think that is the point of view that people typically take--the one Jared is trying to debunk. I think he has a good point but it doesn't apply as universally as he thinks.

Jared, I would never try to poke holes in your methodology (mostly because I don't understand user testing methodology all that much). Perhaps the problem isn't your selection bias, but mine. Maybe my years at IBM have caused me to think that the situation I cited was common, when perhaps it is uncommon. I thought that sites with long sales cycles that require lots of information would naturally have these kinds of searches, but maybe IBM was quite unusual in that respect. Thanks, Jared, for clarifying your methodology and reinforcing the point you made in the study. It's news to me nad that makes it an even more interesting piece of work.

thanks for article.

i agree with ya dear Mike Moran i think u r totally right thanks