Google External Keyword Tool to Be Eliminated; Should We Care?
By Doug Harrington, August 16, 2013
Google has decided to pull the plug on the much beloved AdWords Keyword Tool. In a curious move, they are replacing it with something quite a bit different, the “AdWords Keyword Planner.” On the surface, it would appear to be an improvement over the old; refreshed UI, advanced keyword geo-segmentation, and some tighter integration into the AdWords platform. Buried in the fine-printed labyrinth of help files, however, are some critical changes that shake things up quite a bit.
Phrase Match Is Dead
The biggest change? No more search volume by match type. Broad match – which includes volumes for search terms that contain your keyword terms in any order, and possibly along with other terms – is dead. Worse – phrase match is also dead, that match-type that indicates volume potential for your keyword terms in the order provided, including with terms appended to the front or back of the user query. Search volume is now exclusively exact-match with the new tool: you get volume only for the terms you research, in the order typed. In addition, you can no longer segment search volume by device. Everything is blended into one “convenient” metric. In a nutshell, the data Google graces us with is now measurably dumber.
That’s not the end of the world, but this change means that our ability to be effective in keyword research requires a more comprehensive, intelligent and intentional approach. Unfortunately, that additional layer of effort results in a diminished return. The dumber the data, the fuzzier the predictability of search becomes. To Google’s credit, it makes sense to paint their data more opaquely in order to make paid search more “fair.” It arguably levels the playing field, and bottom-line earns Google more money. But any self-respecting SEM should take note how this impacts their cozy relationship with the accuracy and reliability of Google search data.
So what else is out there?
Moz recently published an excellent article that examined various third party keyword tools, and how the search volume reported by those tools stacks up to Google. It was an excellent assessment, and sparked my interest into digging a little deeper as to where those numbers actually come from. What I found was very concerning.
Virtually every third-party keyword tool sources their data from Google’s exact-match API. Wordtracker was the only exception, which pulls from the search underdog, dogpile.com. Some third party tools like SEMrush blend Google data with their own proprietary sauce to make some Frankenstein metrics – which is fine – but at the end of the day, it’s just a different shade of Google search data. On the surface, it appears that there is a diverse eco-system of keyword data sources, but the darker reality is that it’s just 50 shades of Google.
And then there’s Bing.
Bing may be the neglected step-child of search, but they still have search reach that demands attention. Bing+Yahoo combined have earned a respectable 29.3% of the US search share. Sure, it may be a weird skewed corner of search, but 5.9 billion searches a month is nothing to scoff at. Google, by contrast continues to dominate 66.7% of search share, or 13.7 billion searches a month. (Comscore May 2013) In my view, leveraging both engines against each other is the best approach to diversify keyword data sources.
Drink the data right from the source. Bing data can be segmented in ways we cannot dream of getting from Mama Google. Bing’s sophisticated data elements include:
- Device Segmentation: Desktop, Smart Phone, Dumb Phone (WAP), Tablet
- Demographics: Age & Gender
- Location (US metro)
- Historic Search Volume (2011-Yesterday) with Day/Week/Month segments.
As with any Microsoft offering, there are notable quirks and limitations. For one, the data is locked up in an API. This data is not available on the fly, although there is an intuitive Excel plugin that makes data-fetching accessible to people outside of IT. Even so, data access requires extra thoughtfulness and patience.
Once you get the Bing data-faucet flowing freely, you can’t help but ponder a critical question – how reliable could it possibly be?
Google Vs. Bing
I was interested in stacking Bing search volume up against Google. I took a high quality sample of related keywords for a specific vertical from our database, and collected search volume from both engines for March 2013. Naturally, results are specific to this vertical, and we would expect variance across industries and segments.
After playing around with different methods of comparison as part of an internal case study, I found something quite fascinating. To boil down the process: I grouped keywords into a specific set of search volume ranges for both engines. What I found was that the distribution of keywords within these “search buckets” had a surprisingly strong correlation to each-other.
More outstanding though, was how “perfectly” Google distributed search volume. This seemed like a good opportunity to make a pretty graph. Take a look!
The Perfect Distribution. Dishonest Data?
Notice how Bing’s distribution is much more natural, while still roughly following the same “perfect” distribution set by Google. Interpret as you will, but to me the implication is that Google heavily weighs search volume in a mysterious but no-doubt intentional and algorithmic way. Shocking? Not really – but it is certainly interesting to see evidence of this in the wild. The most valuable takeaway though, is that Bing defensibly appears to be a more natural data source. Reliable? Maybe. Limited? Absolutely.
Where Google had 9% zero-search volume for this sample, Bing had 28% zero-searches. That is quite a disappointment. However, considering they have less than a third of Google’s search reach, it’s still quite respectable. For what Bing did report search volume for, it appeared to be un-molested by an algorithm with questionable ulterior motives. Honest data deserves your attention too.
What Does This Mean?
There are a lot of conclusions that can be drawn here. The validity of each is open to debate. As search marketers mourn the loss of the Keyword Tool, has the data they so love always been dodgy? Welcome to the bandwagon of uncertainty. Is the data we get now so much worse? We can but wait and see.
In the meantime, our advice is same as it ever was: Use keyword research and keyword research data sources as directional inputs, but apply appropriate skepticism as to their accuracy. Objects in the mirror may be larger (or smaller) than they appear.