Is WordTracker's exact match reasoning flawed?

Google doesn't consider exact matches more relevant

[sanderson499]

There seems to be a disconnect in the logic used by WordTracker to calculate KEI. WT defaults to exact phrase matches because (according to their online help) it provides a more realistic assessment of the number of truly competitive pages indexed in a SE's database. That same help screen states that the vast majority of searchers do not enclose phrases in quotes, but that this is "irrelevant".

That seems reasonable, as long as the SE's results favor pages with phrases that exactly match the searcher's input phrase.

But Google and other SEs return significantly different results when a searcher encloses the phrase in quotes and when she does not. Similarly, the "count" figure so crucial to the KEI calculation differs by orders of magnitude depending on whether the searcher uses quotes or not. The more words in the phrase, of course, the greater the difference.

So the question is this: does it make any sense to target phrases based on WT's "exact match" KEI calculation or should I use the SE's overall count number in the KEI calc? It makes a huge difference in determining which phrases are theoretically the most attractive.

[gmoney]

I find that I usually use exact matches to measure a keywords competition instead of the non exact matches. For example I believe that blue widget (900 exact matches) would be more difficult to achieve a top ranking on than widget blue (50 exact matches) even though they have the same number of non exact matches (47,000). This is because Google gives preferential treatment to the order of the words and the word proximity even if the search query was conducted without quotes (i.e. the SERPs for the two queries are quite different). If you optimized for the search query widget blue then you will be competing with 47,000 other pages but only 50 of them will get the bonus for having the exact order of the words. If you optimized for blue widget then you will still be competing with 47,000 other pages but this time 900 of them get a bonus for having the exact order of the words. By using only the 47,000 number you are essentially neglecting the preference Google gives to word order or proximity. I believe that Google gives a lot of weight to word order and proximity and by using exact matches you can capture this phenomenon in your keyword competition analysis. However, the 47,000 number is important because it represents the probability that some big PageRank monster page will dwarf whatever benefit you get from having the exact word order and exact proximity and their “blue super fuzzy widget” page may take the prize of being the top listing for widget blue. Perhaps a better competition analysis would be to find a way to include all three numbers. Actually you could take your analysis to extreme measures by finding the number of exact matches for “blue * widget”, “widget * blue”, “blue * * widget”, “widget * * blue”, etc. Maybe even look at all the PageRanks of the pages listed in the SERPs. You can go even further with super complex keyword competition analysis by trying to capture more of the variables used by Google to rank pages but I usually keep it simple with exact matches.

[gmoney]

So the question is this: does it make any sense to target phrases based on WT's "exact match" KEI calculation or should I use the SE's overall count number in the KEI calc? – sanderson499

I realize that I rambled on a bit in my post above and didn’t really answer your concise question. Anyway, I am not a big fan of the KEI but if I had to use it then I would use it with the exact matches (see above post for long winded elaboration on my views). I believe the KEI scale that is used to determine a “poor keyword”, “good keyword”, and “excellent keyword” is based on the KEI calculated with exact matches (with quotes). If you calculate the KEI with the “non exact matches” (i.e. without quotes) (what you call SE’s overall count number) then I imagine you would need to make a different KEI scale to determine if the keyword is poor, good, or excellent.

Maybe you could make two separate lists of your keywords and rank one list based on the exact match KEI and rank the other list based on the non-exact match KEI. Then you could select the keywords that performed well in both your KEI lists.

[sanderson499]

Thanks for the insights. I guess what's really bugging me is why Google and other SEs apparently give so little relevancy weight to pages that match phrases exactly as entered, even though the searcher omits quotation marks. Seems to me that most casual surfers type in exactly what they're looking to find. Oh well...

Regarding KEI, I find that to be a useful starting point in targeting search phrases, but it doesn't go far enough. After I sort by "raw" KEI I add a "qualification" score...a number from 1 to 5 that reflects my subjective judgment on the likelihood of those searchers being truly qualified prospects for my clients. So the more ambiguous the phrase (widgets vs blue fuzzy widgets, for example) the lower the qualification score. Then I divide the raw KEI by the square of my Q score and re-sort the list by "QKEI".

Since clients can't afford a bazillion optimized landing pages, this technique does a reasonably good job of separating the wheat from the chaff.

[gmoney]

It would be nice, for a low PageRank website owner such as me, if Google gave more weight to exact matches even if the search was conducted without quotes. However, even though I know about searching with quotes and I don’t consider myself a casual surfer, I still conduct most of my searches without using quotes. Maybe future search engines could allow the searcher to dial in the exact match relevancy weighting factor they want for their searches without quotes. If they did this then I agree that the default relevancy weighting factor for exact matches should be higher than it currently is.

Your qualification score is a good idea. I have been doing something similar on my own without assigning qualification numbers but I like how you put it in a mathematical context.