There are many different reasons that keyword suggestion tools are inaccurate. I will cover what I consider the most important skewing factors for Overture and WordTracker. Getting an understanding of why things get skewed will let you more accurately gauge the true number of searches behind any given keyword:
Overture:
This is the most susceptible to artificial search inflation. Hypothetical: Suppose you decide to compete for the keyword online casino gambling. You check it every day (30 searches a month). But your other 50+ competitors are also checking their ranking daily. That is over 1,500 searches a month done that would never produce a single sale. So if overture claims there are 2,329 searches a month, over half of it could easily be non-sense.
MatchDriver: singulars and plurals are for the most part counted together, so when you look up the term casino and casinos, they show the same number of searches, but really each term contributes a percentage of the traffic to the number you see. There is also a type of quirky broad matching, where overture alphabetizes the word order of some searches and lumps various searches together. You can do search for MatchDriver to get a more detailed idea of what it does.
For WordTracker
This tool gets its data from a wide array of different small engines. It takes the numbers it gets from these, and adjusts this for the sample size of Google. This does not work because they do not have a large enough sample size to be accurate. In other words, if a small SEs get 10 searches of any given keyword, and those have 1/200 the traffic of google, they will have to multiply that 10 by 200 making their prediction 2,000 searches per month, but if two more people would have made more searches on those small SEs, then their number would be 2,400! Not very accurate.
Best Regards,
Nicolas Johnson
Gambling Wages Assitant Affiliate Manager
MSN Messenger nicolas_1875 at hotmail dot com