Statistical Validity Of Wordtracker Data, Can you trust the numbers? Options

[Vertster]

I have relied on Wordtracker data for a long time, using it to both brainstorm possible keyword phrases, as well prioritize them for SEO/PPC. Recently, I have begun to question the usefulness of the predict numbers and the implied importance/popularity of phrases. Running Wordtracker reports on the same keyphrases at 6 month intervals has yielded variances of as much as 500% in predicted traffic and therefor implied importance.

On phrases without predicted values in the thousands (which most are not), can the numbers really be trusted as a guage of optimization priority?

What have you done, if anything, to mitigate this problem?

[cline]

Refrain from making predictions. (IMG:http://www.highrankings.com/forum/style_emoticons/default/smile.gif)

I normally start my SEO clients on PPC to determine exactly what they should be optimized for. With the PPC data in hand one can make good forecasts.

[DanThies]

Scott:

Are you talking about keywords with a Wordtracker count that varies wildly from 1 to 5, or are they varying from 10,000 to 50,000? It's the small numbers, right?

We do monthly tracking on a big list of search terms, but I won't even bother tracking a search term with a count below 50. You can't trust *any* data source with the tiny numbers,

[Jill]

I was looking at my server logs for highrankings last night and noticed some phrases that had a lot of searches. For fun, I'd check them in WT and they would say that they had 0 searches.

It would really be great if we had a tool that was a bit more accurate!

[Bernard]

WordTracker uses statistical analysis on searches performed on metacrawler and dogpile:

http://www.wordtracker.com/articles/predict2001.html

I have not seen any studies demonstrating that metacrawler and dogpile have the same demographics among their users as any of the major search engines. I'm not sure if they have some statistical skewing to account for the different demographics among the majors when they are extrapolating the data (or if they do it on a straight market share basis).

[lisety]

.....so one could conclude that it's not very useful to use Wordtracker?
Lise

[Bernard]

IMO, it can be useful, but the predictions may not be very accurate.

[Steve Sardell]

A simple but time consuming work around is to take the key phrase list you have derived from WT and then run actual searches on G or Y for the actual competition. Nothing is as easy as it first appears (IMG:http://www.highrankings.com/forum/style_emoticons/default/sad.gif)

[OldWelshGuy]

I have said this before, but i am getting good business from search terms that Overture & word tracker say are not searched for. Personally I am a lot happier with an adwords campaign as guide, far more accurate, and although it can cost a bit, at least when it does you are getting qualified traffic.

[Steve Sardell]

so one could conclude that it's not very useful to use Wordtracker

Lisety, WT is a tool and if used carefully can and will give you some ideas of key/phrases you may not have thought about. It is not the cure all for all the phrases, but it is useful (at leat IMHO) To note, it is not the only tool for key/phrase analysis, just the most widely used.

[Scottie]

The thing that gets me about Wordtracker is that they use data from metasearch engines... and I don't know any real people who use metasearch engines anymore.

I do think it's a helpful tool for finding some phrases you wouldn't have thought of otherwise, but I wouldn't toss out phrases that really describe what the site does/sells/promotes simply because Wordtracker doesn't report many searches on them. You might only get 1-5 searches a month for that term, but they are so well targeted, they convert searchers to buyers. (IMG:http://www.highrankings.com/forum/style_emoticons/default/thumbup1.gif)

[Steve Sardell]

I don't know any real people who use metasearch engines anymore

Who uses what search engines would make a good discussion in itself. Perhaps shocking, the majority of folks living in my area use AOL (almost on line). Why? Mainly because they were the first ISP in the local, and old habits for old people are hard to break. Re the metasearch, from what I hear university students quite often still use them, especially DP. Then the question becomes whether they are real people (IMG:http://www.highrankings.com/forum/style_emoticons/default/wink.gif)

[Scottie]

I guess it depends on who you want to measure! Who you are selling to makes a difference, and what engines they use.

<There's another thread with that discussion in it going on as well.>

[Steve Sardell]

I guess it depends on who you want to measure! Who you are selling to makes a difference, and what engines they use.

Agree completely. It is part of the demogs that should not be neglected, although, it is often times overlooked.

[Jill]

A simple but time consuming work around is to take the key phrase list you have derived from WT and then run actual searches on G or Y for the actual competition. Nothing is as easy as it first appears  (IMG:http://www.highrankings.com/forum/style_emoticons/default/sad.gif)

Steve, what would that tell you? Only that people think the phrase is useful to optimze for. It doesn't tell you if people actually search for your keyword phrase.

Wordtracker is still very useful, but you just have to remember that for lower volume phrases, they're just not gonna show up there. This is why it's important to use your own server logs too, as well as run an AdWords campaign.

Jill