KEI is definitely a good metric to look at when doing keyword research, here is a quick definition:

Keyword Effectiveness Index (KEI) attempts to characterize the value of a keyword based upon its popularity and the number of competing websites. Higher KEI values are “better” in that they represent more searches per website.

Typically KEI is the the ratio of the square of the searches upon a particular keyword in a day divided by the number of websites that are listed for that keyword. For example, a keyword that has 100 searches a day and for which Google shows 5000 websites would have a KEI of 2. (100 * 100 / 5000)

We have found that it is hard to use KEI, not be cause of the calculation, but because it is hard to get reliable data from online searches.  The tools that use KEI, such as Keyword Tracker, Wordze.com, and others, do an absolutely great job at collecting useful data from available sources, but the source isn’t Google (which owns 70% of the marketplace).  To illustrate this point, here is a FAQ from Wordze.com

Understanding the search industry data collection methods is very important for search professionals to be able to do their jobs. If you have not already noticed, each keyword research service’s data (search count) is never the same. Services like Google or Yahoo always show much higher numbers for most keywords, but fail to include lower keywords with substantial traffic. In contrast, services like Wordze normally show a lower count, but include more long tail keywords than services like Google or Yahoo keyword services.

The general rule of thumb is to use a paid keyword tool like Wordze along with a free service like Google keywords to guide you in your keyword selection. Google’s search service (Yahoo’s service is now defunct) is built for searching for keywords under Google’s Adwords program. This service includes data from Google’s search results, ad placement on affiliate sites (Adsense search), as well as other 3rd party sites like AOL, YouTube, etc. The problem with this method is Google does not show keywords that have good traffic, but don’t make it into the top percentile of their search data. It also means that their search data is normally over estimated for search engine optimization research. On the other hand, services like Wordze will normally show lower count, but provide long tail terms along with ranking data that Google will never show. This typically has to do with the fact that such services do not have all the data, but a fraction of it.

So, like many other tools and services, we use a product like Wordze.com to help gain insight on a particular question, but rarely does a calculated KEI become more than another input point of data for our consideration as we are putting together strategy.  It helps us see outliers, but we often find other Keyword Difficulty Tools (i.e. SEOMoz.org) and a manual review of the Top 10 of Google to be just as useful as KEI.

Related posts:

1. Why is Keyword Research Critical To SEO?
2. Free Keyword Research Tool
3. Google Webmaster Tools Importance to Webmasters