43 replies to this topic

[bwelford]

I agree with the last few posts. It's too complicated to be able to give quick answers.

For keyword phrases, there also is the question whether the phrase is put in quotes or not.

In quotes, you will get only web pages listed that have exactly that phrase. However I think only a small minority of human searchers will be using quotes around the phrase.

If the individual words within a keyword phrase are not put within quotes, then the search engine report will include all web pages that have any combination of the words. There are many factors that influence the ranking order. But ALL OTHER THINGS BEING EQUAL (LOL), the web pages would be ranked in the following order:
web pages with all the words in the order typed, i.e. the keyword phrase
web pages with all the words in the order typed but with one other word somewhere breaking up the exact order
web pages with all the words in the order typed but with two other words somewhere breaking up the exact order
web pages with some of the words in the order typed
and so on. You get the idea.

Then of course there is the factor that some words in your keyword phrase are stop words and are not taken into account by the search engine ...

Stop. It's giving me a headache.

Barry Welford

[compar]

1. There is really no correlation that I know of between the number of pages returned  and the number of people searching on that search term.

Mel,

This maybe nit picking but I don't think that statement is absolutely correct. After all keyword phrases are simply part of the language. If 600,000 pages show up in a SERP it indicates that at least 600,000 webmasters have used those words on their pages.

Even if the pages aren't optimized for the words the number of pages returned is an indication of the popularity or common use of the words or phrase within the language. As such it is probably an indicator of whether or not people might use these words in a search.

Now I don't think you can evaluate the results against the very precise 7 questions posed by Sharon and Roy. In fact I thought the questions were a little naive. Thats why I suggested question "8.) All of the above.". But in turn you can't say that there is no correlation.

[webgrrl]

Hi all,

I just have to jump in here and thank everyone for sharing their processes and experiences in such depth and detail. It's really been enlightening for me, as I'm at that "study-up-and-do-it-right-the-first-time" point... But I should probably stop reading and start doing!

Anyway, thanks for engaging in this great conversation!
I feel like I'm learning a lot.


- webgrrl

[Scottie]

Welcome to the forum webgrrl!

Learning how to do it right the first time will save a ton of headaches later.

[Mel]

1. There is really no correlation that I know of between the number of pages returned  and the number of people searching on that search term.

Mel,

This maybe nit picking but I don't think that statement is absolutely correct. After all keyword phrases are simply part of the language. If 600,000 pages show up in a SERP it indicates that at least 600,000 webmasters have used those words on their pages.

Even if the pages aren't optimized for the words the number of pages returned is an indication of the popularity or common use of the words or phrase within the language. As such it is probably an indicator of whether or not people might use these words in a search.

Now I don't think you can evaluate the results against the very precise 7 questions posed by Sharon and Roy. In fact I thought the questions were a little naive. Thats why I suggested question "8.) All of the above.". But in turn you can't say that there is no correlation.

Hi Bob
I said what I meant to say.
Just because 100 people choose to build a page that uses a particular phrase does NOT mean that a certain number of people are going to search for it, and even if some do there is no way of predicting how many, therefor as I said no correlation.

[Jill]

Yep, I agree with Mel on that.

Jill

[compar]

Yep, I agree with Mel on that.

This may just be symantics. And I'm not saying that the number of pages returned by a SERP is the best way or even a valid way to select terms to optiimize site for. But if 600,000 site contain the same 2 or 3 words then it seems to me you can't say they would never be used in a search term. I agree you can't say how often. The SERP doesn't suggest this. I agree that there are much better ways of selecting keyword phrases, but I don't think you can say there is no correlation.

Pose the question the other way. If you thought you had a valid search term and you typed it into Google and got a SERP of 5 pages, would you accept that as a correlation or would you go right ahead and optimize your site for the term?

[Mel]

If I had a valid search term with so and so many searchers and I found that there were only five competing pages I would optimize a page just as fast as my fingers could type.

[compar]

If I had a valid search term with so and so many searchers and I found that there were only five competing pages I would optimize a page just as fast as my fingers could type.

Mel,

You are just arguing for the sake of arguing IMHO. Unless you have a page for a very rare or exotic subject a SERP of 5 on a search term would immediately suggest that not many people are using it. Because if there were many people using it more than 5 webmasters would have figured it out and included it on their sites.

The reward of being the best of 6 pages -- your's after you optimize and the original five -- just can not be attractive to anyone.

[compar]

This argument is getting stupid but let me make one last point. If you look at Wordtrackers list of 500 most popular search terms and type the top 50 into Google search box, they are all going to give SERPs of 1 million plus. That is called a correlation.

I just think it is unsupportable to argue that search term usage is unrelated to general usage in the language. Search terms are the searcher's best effort to describe what they are looking for. Web content is webmaster's best attempt to describe the nature and content of their web sites. Both group of people use the same language. So language that is popular or common on web sites will also undoubtedly be popular in search terms. It can't be any other way. There is only on body of language.

If I have a low mileage shiny red car for sale I'm going to say so. And if I want to buy a low mileage shiny red car I'm going to search on some or all of those words.

[Mel]

That may well be, but you do run across niches where you can get repectable traffic with very little competition and its my job to find those.

IMO we have a different definition of correlation. I like this one:

"a statistical relation between two or more variables such that systematic changes in the value of one variable are accompanied by systematic changes in the other "

[compar]

"a statistical relation between two or more variables such that systematic changes in the value of one variable are accompanied by systematic changes in the other "

We both have the same definition. To take my top 500 keyword example further I would also venture to say that if you typed the lowest ranking words on the list into Google you would get a significantly lower total of pages displayed using these words than you did for the fifty top word. This is cause and effect. This is correlation by anybody's definition.

[Haystack]

This is correlation by anybody's definition.

I don't think so. It's more coincidental. Think about a breaking news story. The search volume will be extremely high, but the number of pages referring to it will be next to none.

Then search on a person who's no longer in the news (low search volume but relatively high number of pages indexed) such as James Bulger.

[compar]

I don't think so. It's more coincidental. Think about a breaking news story. The search volume will be extremely high, but the number of pages referring to it will be next to none.

Then search on a person who's no longer in the news (low search volume but relatively high number of pages indexed) such as James Bulger.

But those are simply exceptions that prove the rule. People doing keyword research are not using those type of search terms. They are using search terms that they believe relate to a page they may be working with. Nobody is going to optimize a long term commercial site for the search term "blackout in NYC".

Edited by compar, 15 August 2003 - 05:50 PM.