Is the New GWT Search Query Data Reliable?

[aristotle]

I've been studying the Search Query Data in the "New Look" Webmaster Tools for one of my sites. There are two lists, one for impressions and one for clickthroughs, ordered by percentages. These lists are longer than those in the "OLd Look" version, so they should provide more information.

But there appear to be problems with the data:

First, there are several very long-tail queries showing up on both lists, and I'm certain from other data, as well as from common sense, that these long-tail terms are rarely searched, certainly not nearly enough to be on these lists.

Second, according to the lists, one particular query gets 1% of the impressions, but 16% of the clickthroughs. Again, from other information, I know that this can't be correct.

And yes, I've used the dropdown boxes to narrow the data to web searches only and United States (Google.com) only, so that isn't the explanation.

Has anyone else observed these types of anomalies in their Webmaster Tools search query data?

Thank you

[tedster]

I have noticed something similar in a few cases - especially the longtail query terms not seeming to make sense.

Any term that gets 16% of clickthroughs should be notable no matter what kind of server log analytics are used - do your other sources of data include your own server logs?

[aristotle]

After some further study, I believe that the problem may be that the most recent data displayed for this site is based on a smaller sample of searches than the supposed "last seven days". When I check the results for the full month of April, there aren't any rare long-tail queries in either list (both lists show 100 queries), and the overall results look reasonable. I think the "last seven days" is actually a one or two day period at most, and that this accounts for the anomalies.