35 replies to this topic

[000]

what are the "pros and cons" of each service - are they really accurate...


Can one really factor out the searches that are done only to find site rankings - from the "true" queries...


Also, are different "TYPES" of searchers more geared to one set of search engines -than others...

[DanThies]

None of them can separate out the 'rank checking' or 'bid checking' queries.

Google Adwords seems to be the most accurate source of search counts by far, the total number of searches is so large that rank checking becomes 'background noise.'

Overture's numbers are still pretty good, although some search terms can have their results radically skewed by rank/bid checking. It's free, though, and it's a good place to do at least part of your research.

Wordtracker's numbers are pretty useful for "big ticket" keywords, but we often find very specific search phrases that outrank the more generic version when you get to some of the search terms with smaller counts. Wordtracker is pretty good for finding related search terms, and of course the reporting features are useful.

[schecky]

Dan, What if there was a way to determine the skew. If you check Overture for a specific term not used by another site and then check a month later to see the number of queries on that term wouldn't that indicate, to some degree, the skew by remote automated queries?

[DanThies]

I'm not sure how looking at counts for a search term nobody is using would tell you how much the results for other search terms are skewed.

[compar]

Can one really factor out the searches that are done only to find site rankings - from the "true" queries...

I would guess that the number of quiries by web masters would only account for very small fraction of the activity on any keyword or keyword phrase. In any case why would this insignificant factor be any different between one tool or the other? Overture and WordTracker only report the number of searches. They don't adjust or skew the results in any way.

And even if a number of the reported searches are from web masters that is still a bonafide indication of the popularity of the term. If a web master can think of it the probability is that ordinary mortals will also think of it.

The more interesting question for me is how to best analysis and use the information garnered from Overture and Wordtracker.

[Jill]

And even if a number of the reported searches are from web masters that is still a bonafide indication of the popularity of the term. If a web master can think of it the probability is that ordinary mortals will also think of it.

One would think so, but it's not true.

Webmasters and SEOs think of anything they can think of that might be a good search query and they automatically query the engines for those phrases to see where they rank.

There are hundreds of thousands of phrases that appear to get "hits" or searches, when the only ones actually searching using those phrases are software such as WPG.

It's not a small or insignificant number of these automated queries. It's a ton and it totally skews the results. The phrases are no good to anyone, but the SEO who can show his or her client the wonderful high ranking he/she got for the phrases no one actually searches for.

J

[compar]

One would think so, but it's not true. 

It's not a small or insignificant number of these automated queries.  It's a ton and it totally skews the results.  The phrases are no good to anyone, but the SEO who can show his or her client the wonderful high ranking he/she got for the phrases no one actually searches for.

Can I ask to see your data Jill? You make a ton of assertions but never offer any proof or statistics.

I think you are a very brilliant and knowledgable person, but I don't think you know everything and I'd like to see some proof when you contradict any suggestion I offer.

[Jill]

The statement comes from years of looking at the data.

You can easily look at it yourself and see exactly what I'm talking about. Go to overture and type in whatever you want and look at all the keyword phrases that show up as having a whole bunch of searches.

Then optimize a page for some of them. Then watch as nobody EVER comes to your site, because nobody actually searches for those words. They were simply automated queries.

I've seen it a million times. Why would I make it up?

A good tip off to words that are automated queries are ones that are in the form of First Letter Capitalized.

When you see phrases like this: Search Engine Optimization
As opposed to: search engine optimization

You can bet that there's a very good chance that it's an automated query. I'm not sure if Overture is case sensitive but you can see these at WordTracker. I recommend ignoring all phrases that show up in that format.

Jill

[mcanerin]

I agree. My logs show several cases of queries like:

Canadian Custom T-Shirt Design Companies Guelph

Tell me that's not someone running a Wordtracker (or similar" competition check! Especially since it happens at 12:15 every monday.

[compar]

I agree. My logs show several cases of queries like:

Canadian Custom T-Shirt Design Companies Guelph

Tell me that's not someone running a Wordtracker (or similar" competition check! Especially since it happens at 12:15 every monday.

I guess I'm a little dense on this issue. My question is what is the implication for people doing keyword research?

[DanThies]

Canadian Custom T-Shirt Design Companies Guelph

That's pretty typical of how the skew becomes obvious.

On Wordtracker and Overture, you often see phrases like that showing up with higher counts than a more generic phrase like "Canadian T-Shirt Design Companies."

I don't know where "Guelph" is (somewhere in Ontario, I think) but I'm pretty sure it's a lot smaller than "Canada."

[DanThies]

Bob:

The main implication is that you can't necessarily trust counts on search terms, especially at the lower end of the range. A search term that shows 15,000 queries a month on Overture is probably pretty popular, even if a thousand of those queries are automated. A search term that shows 30 a month might be one guy running automated queries every day at 12:15am.

[compar]

I don't know where "Guelph" is (somewhere in Ontario, I think) but I'm pretty sure it's a lot smaller than "Canada."

Guelph is 60 miles southwest of Toronto. You suggest that the use of Guelph in the keyword phrase is a sure indication that it is skewed. But what if Guelph just happens to be hub of the custom designed T shirt business in Canada?

On the matter of the skew -- and this may be a bad example because of the geographic location -- but how could the artifical popularity of this phrase cause anybody any trouble unless they also are in the custom designed T shirt business in Guelph?

[Scottie]

It helps illustrate that there are a lot of automated queries out there and that any search reporting tool will include an unknown number of automated queries as opposed to searches being done by actual people.

[compar]

Bob:

The main implication is that you can't necessarily trust counts on search terms, especially at the lower end of the range. A search term that shows 15,000 queries a month on Overture is probably pretty popular, even if a thousand of those queries are automated. A search term that shows 30 a month might be one guy running automated queries every day at 12:15am.

Of course that is correct and it was the point I tried to make earlier when Jill told me I was all wet. Even if the count of popular phrases is impacted by web masters this can stilll be argued as an indication of their popularity.

I don't think anybody is looking for terms only searched on 30 times a month so the possible skew on this end doesn't matter, does it?