I recently reluctantly renewed my subscription to Wordtracker, a keyword research tool and database. Why was I so reluctant? (And why am I considering asking for a refund?)
Wordtracker provides poor documentation and regularly switches the formulas they use for data they offer without notifying its customers.
Wordtracker Changes the Meaning of “Searches”
In 2009 Wordtracker completely changed the way they calculate the data they provide under the column labeled “Searches”. How they get away with redefining “searches” is beyond me. I contributed to a Wordtracker support discussion about this problem, but somebody else summed it up better. “The fact that a so-called provider of data like Wordtracker does not clearly and expressly explain two critical factors relating to their data is appalling.”
One day “searches” means one thing. The next day it means something else. I got no apology from Wordtracker. I had to eat crow and throw out a few mea culpas to my clients when I realized that Wordtracker had pulled the rug out from under me. Boo. But it gets worse.
Wordtracker Changes KEI Formula
My $329/yr subscription just expired so I renewed. It had been a while since I used the service and Wordtracker neglected to tell me that since the last time I’d used their service they changed the Keyword Effectiveness Index (KEI) formula. I thought I was buying one thing and Wordtracker delivered another.
KEI used to be a great metric to find low hanging fruit or “keyword gems in the rough” if you will. But the metric is now useless for that. Click the image above for a larger version. You can see that the most generic, high traffic, high competition phrases now have the highest “KEI” – and yes, “KEI” should always appear in quotes from now on until… well, maybe forever.
Though they offer great verbiage about what an improvement it is, Wordtracker’s new “KEI” borders on meaningless. Surely somebody at Wordtracker should know that when you combine data inconsistencies with poor communication and terrible documentation, usability will suffer. Apparently the usability and branding experts at Wordtracker haven’t been speaking up.
Comparison of KEI Formulas
Wondering how to determine KEI? Me too.
Typically KEI is the the ratio of the square of the searches upon a particular keyword in a day divided by the number of websites that are listed for that keyword. For example, a keyword that has 100 searches a day and for which Google shows 5000 websites would have a KEI of 2. (100 * 100 / 5000)
– web1marketing.com
So they’re saying: KEI = (daily searches)^2 / Search Engine (SE) listings
Suppose the number of searches for a keyword is 486 per month and Google displays 214,234 results for that keyword. Then the ratio between the popularity and competitiveness for that keyword is 486 divided by 214,234. In this case, the KEI 0.002.
-searchenginepromotionhelp.com
For that one, the formula is: KEI = monthly searches / SE listings
Suppose the number of searches for a keyword is 821 per day and Google displays 224,234 results (pages) for that keyword. Then the ratio between the popularity and competitiveness for that keyword is: 224,234 divided by 821. In this case, the KEI is 273.
-bestpracticemarketing.com
And those jokers say that: KEI = daily searches / SE listings
Those are the first three definitions I found. I’ll bet there are more. Clearly the jury’s out on KEI. But while contradiction abounds, there’s a common thread in defining KEI. It has always related to the quantity of searches and the number of search engine listings.
So what’s the new Wordtracker definition for KEI?
Maybe we should start with the old Wordtracker definition of KEI
KEI compares the Count result with the number of Competing Web pages
Yep. That seems to be in line with what everybody else says about KEI. In case you were wondering, “count” is, “The number of times the search phrase has been used in Wordtracker’s partner search engines.” And “competing” means, “The number of Web pages the search engine says it has in its index that match the search phrase.” So more specifically the old Wordtracker formula for KEI was
KEI = (daily searches)^2 / Search Engine (SE) listings
Here’s the new Wordtracker KEI definition
KEI = (Searches ^ 2) / In Anchor
Is that searches per day? Per month? Who knows? The only other information Wordtracker provides on its data results pages about its new KEI equation is
KEI compares the number of times a keyword has been searched for with competition (the number of pages that contain the exact keyword phrase within at least one of its incoming links, known as ‘All in Anchor’).
Does the “In Anchor” include only external pages? Or will a page with an internal “In Anchor” link make the cut too? Tough to say. Wordtracker regularly defines things their own way. While I’ll not poo poo innovation, I take umbrage with my data providers when they skirt industry norms. If Google defines a metric a certain way, clearly it is beneficial to follow the leader. Note to Wordtracker: Don’t confuse your users by regularly creating new definitions for established industry terms. Your poor usability is a disservice to your paying customers.
In Anchor And Title IAAT
Wordtracker founder and CTO Mike Mindel says
‘In Anchor and Title’ is a count of the number of pages for which the keyword appears in both the title tag and the anchor text of at least one backlink to the page (not domain).
Understandably this metric is used to help identify serious competitors. But Google measures parts of this metric differently. Back to Mike Mindel
There are two reasons why [Wordtracker] and Google show different numbers of links for seemingly similar searches. The first is that the [Wordtracker] In Anchor metric shows a count of external anchor text (from other websites), whereas Google includes internal anchor text as well (from within a website).
‹rant› If internal In Anchor links are good enough for Google they should be a sufficient metric for Wordtracker. Wordtracker tries to sell you on why its better to use their more specific metric, but aren’t all search engine optimizers essentially trying to play Google’s game?! Why wouldn’t Wordtracker emulate Google metrics as much as possible? Clearly they’re meaningful. Something more specific isn’t always better. Furthermore, why would you use the same terminology to discuss two separate things? ‹/rant›
Wordtracker’s Mr. Midel goes on to say,
The second reason is that Google’s AllInAnchor returns broad matches by default (the words mcdonalds, nutrition, and facts in any order), whereas Wordtracker uses the In Anchor phrase match count (mcdonalds nutrition facts somewhere within the anchor text).
(See previous ‹rant› .) Mike Mindel continues,
I hope you can see now that bigger numbers clearly do not mean better numbers.
Well, Mike, I hope you can see now that I’m not sold on your new (bigger) KEIs being better than the older, smaller KEI figures. And doesn’t Wordtracker try to sell us on bigger numbers being better? (See next paragraph.) Now I’m confused(er).
Back to Low Hanging Fruit
This new KEI formula doesn’t do much to help SEOs find keyword phrases with low competition and reasonably high traffic. It’s more tailored to high traffic phrases. Mark Nunney of Wordtracker says,
“KEI squares Searches because otherwise if both Searches and Competition (whatever metric is used for this) go up at the same rate then the KEI value remains the same and that will not take into account the increased opportunity that more Searches offers.”
I don’t know… I always thought that popularity proved only popularity itself. (Think: MC Hammer.) I also always thought that KEI was to represent some notion of ROI. Big returns aren’t valuable if the investment doesn’t make sense. Even my largest clients benefit from low hanging fruit and the small investments required to conquer them. Just because one has the deep pockets necessary to go after high traffic keywords doesn’t mean that it’s the most effective path. Mull it over. Easy pickings are more valuable to me than the garbage these new Wordtracker metrics provide.
Wordtracker Alternatives
I wrote an email to Wordtracker explaining that I want a refund. But I haven’t sent it yet. I looked for wordtracker replacements. There are a few that are too expensive for me to even consider. (We’re talking $1000 per client per year.) But I found a few tools that provide good data. They are:
https://adwords.google.com/select/KeywordToolExternal
http://www.google.com/insights/search/#
I’ve also heard decent things about marketsamurai.com/ but I haven’t tried them yet, so no endorsement or link out.
The sad truth is that I think I might get enough value from Wordtracker to warrant sticking with it. After all, I can dump all of the data to CSV and make my own versions of KEI to get the data I want. That’s nice, but that’s not the point.
Wordtracker sucks. They keep changing the definitions without notifying customers which causes Wordtracker’s usability to suffer. I am searching for Wordtracker alternatives. Let me know when you find a good one. I’m willing to pay for a wordtracker replacement.
Dan Dreifort consults on SEO and usability for companies large and small. He whines a lot on this blog. Sorry.
Dan Dreifort 
Got a reply from Mal over at wordtracker. It’s quoted below for your convenience. Original reply here:
http://getsatisfaction.com/wordtracker/topics/why_has_kei_equation_changed_whats_with_perpetually_poor_wordtracker_documentation#reply_3417967
Hi Dan,
Thanks for taking the time to write. I’m sorry you’re so annoyed by this, but I hope I can clarifiy our position for you.
In regard to your first question here, It’s important to recognise that the KEI was designed some ten years ago.In the meantime there have been vast changes not only in the Internet generally but also in SEO practices. As a result of this, KEI, while still useful, has changed its spots a little. It’s no longer safe to say that a KEI of X or Y is good or bad, as this metric is going to vary from niche to niche – each niche is going to have a different share of search and a different crowd size competing for that search, so it’s much more useful to look at the numbers in relative terms rather than the raw figures.
I see from your screenshot that you’re searching data from the Google AdWords API that Wordtracker is currently testing. As Google’s search counts are often enormous (particularly in comparison to other independent keyword research tools’ results), this will inflate KEI (KEI goes up when there’s a higher search count and down when there’s higher competition). However, looking more closely at the list, even though there’s a KEI in the tens of millions (which at one point would have indicated a keyword a webmaster could retire on), the In Anchor and Title metric tells us there are over fifty thousand pages directly optimized for that keyword.
The Google AdWords data is as I’m sure you’re aware, a markedly different dataset to the Wordtracker metacrawler data and does need to be treated a bit differently with regard to the metrics that are produced. As I mentioned earlier, it’s almost invariably more useful to look at the figures presented as relatives rather than as absolutes.
You know what though? You’ve got a good point about the documentation. I was looking at your blog post at your dandreifort site, and realising that while all the KEI information is there in Mike’s article, we could do better in terms of making it more obvious, and we could also do better with our communication about the Google test.
Again, thanks for getting in touch, Dan – I’m sorry you’ve been brought to such frustration, but I’d be happy to give you a call if you’d like to discuss this further – either drop us a line at support@wordtracker.com, or to me directly at mal@ and we can speak.
All the best,
Mal
nice tutorial brother, thanks a lot
so did you get new alternatives for wordtracker? or shall we build one? 🙂
I’ve largely been making due with free tools like
…and a few others. …Relying on Wordtracker à la carte (monthly) as a bonus for larger research projects, e.g. new clients and realignments, as they come along.
I find that paying them annually is no longer worth it.
hi, I am new to keyword research and I really want to get the right tool for the job. Do you know other tools where I could get accurate competition numbers or the number of results on google could be used as a basis for competition. I am a programmer and I would like to create a program for personal use. Thanks!
@aivan
The standards I’d use to determine competition, in order of increasing weight, are:
— Google results
— Google results in quotes
— Google results in title
— Google results in anchor (any) and title
— Google results in anchor (external) and title
So Google is the way to go. If you can reliably glean the correct data from Google, (without getting blacklisted for violation their AUP/TOS,) then that’s the way to go. I’d be happy to advise more. Send me a PM for more info.
Hi Dan,
Thank you for your detailed reply. I wanted to PM you but I don’t know where. I don’t know if you could see my email on your database. please message me on that email.
Let’s talk in email-land!
Hello Dan,
This is an old posting you have there, but I found it today and want to add that there is another huge flaw with the Wordtracker KEI (which their support team doesn’t want to answer). It gives a KEI of zero for words with zero competition, because the formula divides by zero, when in fact, these keywords could be your best pearls.