Most of the time, beginners are advised to do keyword search using Google adword tool. While that is a good start, I find that the tool is useless for longtail keyword that beginners should target. What I am currently using now is a different approach.
This is how I do keyword research. It is a variation of the common strategy that IM folksuses. When analysing keyword competition, they look out for an ezine or hubpage or other web 2.0 properties on the first page of the SERP, because that is a strong indication that the keyword is easy to beat. However, this process is tedious and is very dependent on luck. You just need to try a lot of keywords to see which one has the result that contains these Web 2.0 in the first page.
What I do is the reverse. I first analyse which ezine article is already getting lots of views in a short time by using the following search string:
site:ezinearticles.com “This article has been viewed″
The first part allows you to see the no of views directly on the SERP so saving you the trouble of clicking into individual articles. The second part is your niche. The third part will filter the articles according to date.
For me, I look for articles that have been written in the past 3-4 months and see which ones have a couple of hundred views. This tells me that the market has interest in the niche.
Then I look at the post article and try to deduce what are the keywords that the article is ranking for. I then do a search using the keyword that I have deduced. If I see that article in the first page for the keyword, it means that the keyword is easy to rank for.
The reason I look only at the past few months is to take away the age factor. If an article written in April has gather ranking and generate hundreds of views within 1-2 months, it means we can probably achieve the same result by either using an EDN or creating a page on your current site and point several Web 2.0 properties to it.
As an example, this is what I found just 15 minutes ago using the diet niche
Submitted on 13 April 2010,
Ranking for: 1300 Calories Diet Menu
I have tried this method a couple of times and so far, the results are what I have expected. However, I need to experiment for a longer time to draw any meaningful conclusion. In fact, I have all these data in my spreadsheet. Once I have enough sample, I hope to derive some formula such that I can estimate any monthly traffic based on these few statistics that been mentioned.
If this method works, it will be interesting because I can use the same approach to data mine all article directories like ehow, associated content etc. In fact, with some alterations to the search string, you can identify a lot of big seasonal money makers : )