I posted the follow on the Warrior Forum...Thought I should post it here, as well:

My own keyword program PowerBomb, which I took off the market years ago due to faulty third-party results, used a more sophisticated formula than any other keyword research tool.

But first, you can't use a straight ratio, it's too inaccurate. For example:
1/1 = 1 (KEI)
100000/100000= 1 (KEI)

Both of these give us the same "expectation", however they are two totally different situations.

1/1 is very easy, however, there's only one search per month. Hardly worth going after, whereas the second example would be pretty hard to crack, but offers a good amount of traffic.

Again, these are two very different situations that a straight ratio tells us are the same.

Next would be KEI, which "dampers" the traffic numbers. This is better, but still not accurate enough, as it leaves out the VALUE.

Therefore, we need to factor in the VALUE of the keyword phrase to get the most accurate info. Value is calculated by the amount paid for the keyword in a Pay Per Click search engine.

I won't give away my exact formula, but it would be something like:
(Traffic X Damper) x (Traffic X Competition) X Value = Power of Keyword

Now here's the problem and again why I quit selling my PowerBomb: Accurate third party stats became impossible to get. I you have crap going in, you'll have crap coming out and it doesn't matter how good the formula is.

GoTo (Overture) used to give accurate info, including cost per bid and traffic. The info around today just isn't as accurate as it needs to be.

Having said this, IMO, good keyword research isn't finding a keyword phrase or two that are "good"...It's about finding a niche that has scores and scores of good keywords and phrases that can be mixed and matched in different combos.

Another thing, keyword research is really only accurate for those using SEO/SEM. If you're using other methods of marketing, keyword research isn't needed to a great degree. You may want to research the broad, general niche. But to drill down is a waste of time.

Search engines are like creating a shopping list. But while folks shop for items on their list, they also see (and buy) many other things.

If you have a product/article that will appeal to folks if they are exposed to it, keywords don't matter. Just as having milk and eggs on your shopping list, you see a box of Twinkies and buy them, even though they weren't on your list.

Another point: Research is just that...Exploring things that happened in the PAST. With a little insight and creativity, you can predict some keywords that have zero demand now, that could be very popular in the future.

I've given the example of hurricanes in the past. Names for hurricanes are put out years beforehand. Most of these names will be meaningless. But if one hits a major city, this particular keyword will become hugely popular. Here's an example of a keyword phrase that would have poor traffic numbers before hand, but will greatly increase, and any research is useless.

Here's another example of what I call "keyword speculation"...American Idol. Every year there's 25 unknown "keywords" at the beginning of the season that get zero traffic and have zero competition. After about a month, the remaining singers will create huge search totals.

If you're into keywords and research, you may as well take the next steps:

- Understand that today's stats aren't very good.

- You only need to research SE keyword stats if you're using SEO/SEM (and maybe PPC)

- A good "keyword" doesn't do much good for SEO. You need a niche with a lot of good keywords

- Research is history. See if you can't "speculate" about what keywords may be hot in the future. Get a head start on other marketers. Find niche magazines that discuss products that are going to be released soon.