63 replies to this topic

[rtysmith]

Jill - I'm not saying your biased, but your testimonial does appear on KD and they appear to be an advertiser of your newsletter.

Since reading your newsletter and seeing the ad, I've looked at KD (we're a long term WT user) and sent in an email asking what the benefits of KD are over WT as well as Trellian's other products. As of yet, I have received no response - I've been waiting about a week (they may just all be at the SES show).

They claim the KD database is 78x the size of WT, however with the above results it doesn't make sense.

I thought there would be some correlation between database size and results, but its all over the board. CSV results are below.

service,database size,tiger woods,tiger woods wife,tiger woods girlfriend,tiger woods golf
KD,26000000000,418842,14616,12632,5610
WT,330000000,3443,565,236,110
ratio,78.78787879,121.650305,25.86902655,53.52542373,51

Overall - it still leaves me lost for which is the right solution - I guess you just pick one and go with it.

[Chris B]

I just started to use kw discovery 2 weeks ago. I ve seen some discrepancies with my actual stats. I can think of 3-4 words off the top of my head that kw discovery will show 100-400 in their database with a count between 2-6, while im getting 15-30 visits a day for these particular keywords and most of them coming from yahoo and msn.

While WT 6 months ago, showed the predict to be closer to what im getting, still off but closer. and i dont expect perfection but would like to see something closer. 2 of my main kws bringing me the most traffic, if i was using kw discovery, i might not have optimized the pages for these words thinking that no one was searching on this. thats why many have said, its always best to doublec check with your own logs.

now on the other hand, my other site im working on, it came up with a lot more keywords than WT. so my own personal solution for the time being, i will use both.

my only issue with kw discover is my projects kept changing on me, very frustrating to add 600 keywords to the wrong project...GRRR... and a few times, i had to quit bc the project wouldnt change. i spent way too many hours on this, mainly either starting over or going back and deleting the kws in the wrong project. though they emailed me last night and said this should be fixed

Chris

[AlDugan]

my only issue with kw discover is my projects kept changing on me, very frustrating to add 600 keywords to the wrong project...GRRR... and a few times, i had to quit bc the project wouldnt change. i spent way too many hours on this, mainly either starting over or going back and deleting the kws in the wrong project. though they emailed me last night and said this should be fixed smile.gif

I was having this problem too... it was very annoying.

The fact about keyword research tool is however, that you can't expect them to be totally accurate... if you want to know how many people are searching for a particular term on Yahoo or Google then you have to get that information directly from them... and the best way to do it is to advertise for those terms. You don't have to be in position 1, but if you want the accurate impressions a keyword gets then you have to bid for at least page 1... and I don't really see the logic in the argument that you can't afford to do that (I believe the original poster mentioned this) because if you are trying to rank organically for a given term then you must be assuming it is a term that will convert traffic... thus any PPC traffic you get from that same term should convert as well.

Now I'm not saying there aren't any terms that are too expensive to bid for page 1 but for the most part if you're picking the right keywords, then they should also be profitable on a PPC campaign...

...hope I'm making sense, I'm kind of rambling again...

[Scottie]

Well, like I said in the beginning of this thread, there are no absolutely accurate keyword tools. They can give you an idea of the relative popularity of different versions of terms, but they can't predict how many referrals you will get for any specific phrase.

If you want to talk about how inaccurate they are, we can fill up pages and pages because they do not cover every search and they really can't predict what your results will be.

Top that off with the fact that everyone who searches for a term may not in fact visit your page (especially if there is something about the listing that turns them off) and there are too many variables to count.

Keyword research tools are just for research. They are great at giving you an idea of what you want to target but they can't possibly tell you how ranking #1 for each term will affect your traffic.

And YES- that project-switching thing at Keyword Discovery makes me nuts. I've started logging out and back in when switching projects.

[Chris B]

And YES- that project-switching thing at Keyword Discovery makes me nuts.  I've started logging out and back in when switching projects.

Scottie and Al, glad to see im not the only one, i thought i was going nuts not to mention my poor desk was taking a beating!

[David1221]

David, what was the reaction of the manager?

In regards to this, the manager simply came up with a "by the book excuse" and that was "oh... well we get our results from 180 search engines and WT uses Dogpile and Metacrawler" In other words what he was trying to say is that they get their results differently...... but I still don't feel that only 233 searches per year is a justifed number considering KD tackles overture search engine results as well...... and if overture comes up with 853 per month (off the top of my head) then i think KD would have to have a larger number for a YEAR of searches.....

You guys all have brilliant input and i thank you for it...... the whole idea is to point out that these tools don't support each other in some cases and that it can make selecting keywords difficult.... For now I think my strategy will remain the same and that will be to use WT and lets say it returns a high result for a keyword, I will look in overture to make sure that it is close enough to the figure of WT (i think a 40% difference between WT and overture is reasonable for a keyword, meaning it must have some genuine traffic).

Goodluck to you all

[davidfromraleigh]

Keyword research tools are just for research.  They are great at giving you an idea of what you want to target but they can't possibly tell you how ranking #1 for each term will affect your traffic.

And YES- that project-switching thing at Keyword Discovery makes me nuts. I've started logging out and back in when switching projects.

Amen to both of those comments Scottie.

David

[projectphp]

there are no absolutely accurate keyword tools

Ahem. What about the i s o s 100% accurate keyword research tool?

Look at our totals for online website marketing, and tell me that isn't 100% accurate!

I mean come on Scottie, how quickly we forget!

Edited by projectphp, 28 February 2006 - 07:41 PM.

[projectphp]

the whole idea is to point out that these tools don't support each other in some cases and that it can make selecting keywords difficult

Not really. IMHO, keyword research isn't about absolutes, it is about starting a marathon with a 2 km advantage. You still got 40km to go, but what an advantage!

The trick of KR is to get a feel for what you should be doing, and then reassess down the track. Not to believe the results are the final indicator.

Because totals change. imagine Condoleesa Rice (shudder shudder shudder) is the Republican nominee in 2008 (shudder shudder shudder). Her search totals will boom, probably ten minutes after she is confirmed (shudder shudder shudder).

To assume keyword totals are static, and that you need anywhere close to 100% accuracy to make a decision, are myths IMHO. What you need is an indication, and then constant reassessment of what is going on, how and why.

[FP Guy]

Well, like I said in the beginning of this thread,  there are no absolutely accurate keyword tools. They can give you an idea of the relative popularity of different versions of terms, but they can't predict how many referrals you will get for any specific phrase.

I know that the results vary and my solution to this is to compare the results from my own research, a software I use, Wordtracker, and Keyword Discovery. If it looks fairly decent in all of them then I will go with it.

Many times I'll find an outstanding number in one, and a virtually non-existant one in the other. (Keyword Discovery is usually the one that spits out non-existant)

I'm glad someone started this topic, because I've always wanted to know what others thought.

And yes, I came across the same problem with keeping my keywords straight in the projects.

[Jill]

You guys need to all complain to Trellian about the project switching thing, because we've been working with them for months to resolve it, and they seem to keep saying that we were the only ones complaining!

If everyone else complains, perhaps it will get fixed sooner.

[Chris B]

Jill,

They told me 2 days ago its been fixed. I will find out tonight.


Chris

[kdmaster]

I think I can clear up some of these misunderstandings and issues raised in this forum.


1. Word Tracker Predicted totals vs. Keyword Discovery totals
---------------------------------------------------------------------------
The WordTracker predicted daily total is an estimate based on wordtracker's data sample. It is an extrapolated figure based on what WordTracker believe is the total number of internet searches.

So if WordTracker sees 1 click per day on a given term, and assuming they have a sample of say 0.25% of the total internet searches, they would report that one click as 400 daily searches.

The KeywordDiscovery total is the TOTAL number of searches in the KD database. That is, the total number of ACTUAL searches recorded on that term, not a made up estimate.

This means that you can not compare these two totals as the KD total will almost always be lower.


2. Discrepancy on some results
---------------------------------------------------------
It is true that there can be some huge differences in the reported search totals on some terms. There are 4 main reasons for this:

a). The main reason for large extremes are parked domains. Parked domains account for a very large chunk of search egines's revenues and the results they display are classed as searches. For example, if you go to this site: www.kazakov.info, sedo (the domain parking provider) sends a search request to Google to retrieve and display the search results for the term "medical software". This counts as a search every time anyone simply visits the page. This inflates the number of searches for that term and this is often reflected in the KD data (mostly on popular search phrases).

This is why you see searches for things like "online website marketing" which real users may not search for. They may not search for it implicitly, but they will still see the results for this query if they happen to visit an expired SEO related site.

WordTracker does not collect data from Google or any other search engine that powers the parked domains, so their data does not reflect this source.

We have decided to include this data because these results are classed as searches by the engines and if you are advertising on the Google search network, you will be charged for clicks generated from parked domains.

WordTracker use Dogpile and Metacrawler for their data and these two engines will a have a more specific audience. The less web savvy users will tend to use Google, Yahoo, MSN or what ever is presented as the default search in their browser. This can again cause a discrepancy as KeywordDiscovery includes a search sample from engines like Google, Yahoo, MSN, Aol etc.

c) Lastly, searches generated by non-english speaking engines may also generate a large discrepancy as WordTracker does not include regional engine data.


3. Project Management Issues
----------------------------------------------------------
I believe that the bug with Keyword Discovery projects has been resolved two days ago. If not please email me.

As a side note, if you manage a large number of keywords in your projects, it is much easier and more flexible to do this on your PC.
The SEO Toolkit is integrated with KeywordDiscovery and includes a Keyword Manager that lets you import, manipulate and manage multiple keyword projects locally. You can also make you own backups. (To download, click the SEO Software link on KeywordDiscovery.com).


4. Keyword Discovery Support
---------------------------------------------
There was a complaint raised in this forum that the KeywordDiscovery support took 20 minutes to come back with an answer. As the CEO of Trellian, such comments always concern me. In this case the issue had to be escalated, as it involved a comment on a competitor. Also after reviewing the logs, the 20 minutes was in fact closer to 5, however IF ANYONE has any support issues with KD, please email me directly.

Hope that helps

Edited by kdmaster, 02 March 2006 - 07:42 AM.

[FP Guy]

So if WordTracker sees 1 click per day on a given term, and assuming they have a sample of say 0.25% of the total internet searches, they would report that one click as 400 daily searches.

This is a good thing to know.

WordTracker does not collect data from Google or any other search engine that powers the parked domains, so their data does not reflect this source.

We have decided to include this data because these results are classed as searches by the engines and if you are advertising on the Google search network, you will be charged for clicks generated from parked domains.

Hmmm, which way would closer relate to human searches?

WordTracker use Dogpile and Metacrawler for their data and these two engines will a have a more specific audience. The less web savvy users will tend to use Google, Yahoo, MSN or what ever is presented as the default search in their browser.

I use Google, just because of the accuracy of the results and with Google taking over 46% of the searches (Neilson NetRatings November 2005) wouldn't it be better to go with what people are using instead of Dogpile and Metacrawler results? I wouldn't say that people who use Google are less web saavy though.

I appreciate you giving the time to put your input here kdmaster.

[kdmaster]

I use Google, just because of the accuracy of the results and with Google taking over 46% of the searches (Neilson NetRatings November 2005) wouldn't it be better to go with what people are using instead of Dogpile and Metacrawler results? I wouldn't say that people who use Google are less web saavy though.

Google was perhaps not the best example, more the AOL, MSN .... But I agree, we would love to get access to full Google search logs to base the KeywordDiscovery database on. Obviously that is never going to happen, so we have to make do with the current technology that scrapes the google search data from ISP logs.

ISP logs are a great source of quality data, but as was pointed out, it is limited to the choice of ISP and the type of audience they have. Still, when aggregated with the rest of the data provided by other (mostly second tier) engines, it provides a good spread of keywords and phrases.