The imperfect logic of using Overture + Extension to estimate traffic

Is "domain tasting/testing" the gold standard for projecting traffic?

[Webwork]

Using Overture's Keyword Tool to estimate a domain's traffic level is a weak tool, at best, for projecting traffic . . but that have never stopped the masses - including myself - from making reference to such data.

The keyword tool simply indicates the number of times someone on the Overture network has entered a full domain into a network search box.

Riddle me this: What, exactly, is the logical connection between people who search by entering a domain name into a search box and people who type a domain name into a browser address bar? I mean, aren't people who type in domains into a search box a different breed than people who type in domains into a browser address bar? So what's the connection? Why should one behavior serve to predict the other with any accuracy?

I'll give you this much: The volume of people entering a domain in a search box is some evidence of the fact that people believe such a domain/website exists. Therefore, people entering a domain in a search box might indicate the likelihood that other people - having the same belief - will take the shortcut of simply typing the domain into a browser.

But really, why search a domain in a search engine? What's with that? Should the actions of such minds/people be employed for other purposes?

This doesn't even begin to address the disconnect between Overture numbers and actual type-ins.

Let's also overlook seasonal variations in a domain name's traffic. For example: Allow me to bid on your skiing domain based upon May's stats, okay?

So, now that I've disparaged one approach to estimating a domains traffic allow me to ask:

What do you employ to project or estimate domain type-in traffic?

In case you think I'm donning my clueless cap - yes, I understand one can seek out the stats for a domain, but that assumes it's a) an existing domain; and, 2) that someone is tracking stats . . and yes, one can test an existing domain . .

However, I'm considering speculative registrations - so how do I test before registering?

My old stand-by, especially when considering pursing 2 word domains as new registrations or aftermarket purchases, was a Google exact match search: For example "business directory" [google.com]

[chatcher]

I also have always been skeptical of using Overture numbers to predict traffic. Of course, there are a number of people who do it and have done very, very well. So much so that just about any expiring name with good numbers will be very pricy to obtain.

I think it is more reasonable to focus on niches in which an intimate knowledge of how to monetize the topic gives you an advantage, and the potential returns from less contested names become worthwhile.

For available names you can always register, test, and delete (at some registrars). But available names with much traffic are usually snapped up by someone with access to DNS traffic (monitoring popularity of nonexistent domains).

[cerebrum]

These people are used to search engines and they type virtually anything into these search engines. For example, they found webmasterworld.com in a search query on google, they remember that first page SERPs contains a useful site called webmasterworld.com. Next, they come and type webmasterworld in search box and then will go to the site. Many people are confused with com,net,org,de,biz,info,mobi,crap and trust only google and enters webmasterworld in the search box instead of direct type-in.