(14)
Michael VanDeMar wrote an interesting post on his blog about the uselessness of the Google Keyword tool for SEO. He explains that while we reported that the tool is showing keyword numbers, it still isn't helpful. His rationale is that the numbers refer to PPC search behavior only, not overall Google.com search behavior, and then he explains how past research has given him this perception. He writes:
For instance, for [birthday poems] the tool gave a number of 27,100 (which would be an average of 903 searches per day) and bidding on that keyword for 3 days gave me 2,411 impressions (or 803 impressions per day). This is fine and dandy if I am only concerned about getting traffic from AdWords, of course. The thing is, if I rely on this data for my SEO efforts I will at best be most likely wasting my time.
Interesting analysis. Jill Whalen admits that the numbers seem rather high but it now makes sense because they use the content network in their numbers.
Forum discussion continues at Sphinn.
Comments:
James Norris
07/16/2008 03:38 pm
Damn it. I hadn't looked at the tool since I received a SEroundtable twitter about the Adword tools update. What would you guys suggest??? I used to use Goodkeywords (god rest its sole) and now use Wordtracker. (Wordtracker UK results are fairly poor with no information where they gather their data from!).
Eric Itzkowitz
07/16/2008 04:18 pm
No keyword tool is completely accurate. I only use them for discovery and trend spotting. IMHO, keyword research tools are merely one tool in the good ole SEO Toolbelt I can use to help me make a more educated decision on where to focus my marketing energy.
Michael Martinez
07/16/2008 04:52 pm
If you look at the numbers for competitive queries you should see evidence of rank-checking (consistency in numbers across a lot of similar phrases). However, there is no such thing as "PPC search behavior only". The idea that Google is somehow distinguishing between queries for PPC and queries for organic SEO is absolute nonsense. The test he describes could have been flawed in many ways (and probably was, as most SEO tests are usually very flawed). People should not ignore these numbers as they do provide a sense of scale you won't find anywhere else.
Tyson
07/16/2008 05:30 pm
Content network? How is that possible? I believe they use the search network, but not the content network.
Michael VanDeMar
07/16/2008 05:44 pm
Tyson, you are correct. Jill must have misread what I posted. I never said that they used the content network... however, they might by. I was mistaken in thinking that you cannot filter out the keyword tool itself by location. I somehow missed the fact that it was defaulting to US traffic estimates. This means that the numbers were further off than I was calculating, and that the content network might actually be the bulk of the difference.
Ryan Adami
07/16/2008 05:49 pm
I strongly disagree! Please view my comment on the above mentioned post, and my full response on the nonBot blog. (http://www.nonbot.com/blog/index.php/2008/google-keyword-tool-update/) Simply put: PPC ad impressions are searches!
Tim
07/16/2008 06:30 pm
I agree with Micheal in that this tool is not perfect however, I greatly disagree with the test he has run. Three days of bidding on a term that he could have easily out spent his budget on. If his budget was spent during that day then he would not have accurate data because he would have missed out on impressions. Also, the number he uses is an average over at least a year. He used three days. Did he figure out if the difference he saw was even statistically significant. I would have to guess no. you have to take into account the seasonality of searches. Also was he on the first page for every search? That being said, I think the keyword tool numbers are a little inflated. Google wants to put it's best foot forward. They want people to advertise, these numbers likely include ZERO filtering as in the way of robot searches and other automated inquiries. Why, because if the numbers are higher people are more likely to spend on terms. Also, in competitive industries like Internet Marketing, Real Estate and Others, there are so many rankings checkers and keyword researchers that I am sure it inflates the numbers of real people searching for these terms. It is a great tool, and might even be better than anything out there now, but it is not perfect.
jeremy
07/16/2008 10:10 pm
In terms of paid search, my data suggests the tool is about 95% accurate. Far from a complete analysis but a good starting point. It actually underestimated the data in this case. http://www.ppcdiscussions.com/2008/07/how-accurate-is-adwords-keyword-tool.html "When logged into the account the keyword tool indicates that the approximate June search volume was 34,830. That estimate is based on exact match and the correct location targeting. Here's what surprised me - the actual number of impressions that keyword received in June was 36,770. Do a little rounding and the estimate provided by the keyword tool comes out within 5% of the real number."
Vineet Waldia
07/17/2008 05:54 am
In my opinion wordtracker is good tool and may i know which is the best tool for keyword research. http://www.seostep.net.
Vineet Waldia
07/17/2008 05:55 am
May i know which is the best tool for keyword research.. I use keywordtracker and googlesandbox for keyword research http://www.seostep.net
Duarte Canario
07/17/2008 08:37 am
It's not perfect, but for international non-english research, is probably the best solution. If James complained about Wordtracker UK results, imagine their portuguese, polish, or swedish results...
Alex
07/17/2008 01:28 pm
In any case real numbers are better than those graphics...
Justin Seibert
07/17/2008 02:19 pm
You probably can't look at the results, which are general anyway (look for high volume words and tell me if you see anything but zeros at the end), and declare that there are __ # of searches per month on ___. Having said that, keep in mind that Google is still only 2/3 of the search market, so even inflated, the actual number of total searches across engines is higher than what Google reports.
Mahesh
07/25/2008 05:43 am
I am totally disagree with you man. :(