Can you trust your keyword data?

The Holy Grail of Opportunities
June 24, 2017
Show all

Can you trust your keyword data?

As the clever marketers we think we are, there’s no riding into online marketing battle without thorough examination of the available data. Without a well thought out strategy, we’re just wasting time and money on the hope of getting lucky. We have to know our market niche, so we can plan where to attack, where to avoid and which parts to save for later. We also need to learn about emerging trends before anyone else does, so we can capitalize on them swiftly with fewer resources.

However, what if we can’t rely on the keyword data we’re getting? If you’re just using the Google Keyword Planner without enough ad spend on your Adwords account, you most likely have already learned how unreliable the data is.

Note: I also added “designed” to the list within keyword planner, but it got grouped with one of the three shown, highlighting another issue which I’ll address later in this article.

I don’t know about you, but those ranges don’t seem very helpful to me. What we really need is an Adwords account with high enough ad spend, or a different tool which has access to this data.

As comparison, here’s data from keywordkeg:

This is much better now. Theoretically, if we’d take every data point of the last 12 months individually, counted them together and then divided the sum by 12, we should get the exact same search volume that is shown, right?

For “design”, we have (4*165k) + (6*135k) + (2*110k) = 1690k, divided by 12 makes almost 141k. While there’s “only” a discrepancy of 6000 searches, it highlights another issue we’ll address now.

Google Keyword Planner’s obscuring search volume

The before mentioned discrepancy of 6000 searches per month between the 12 month average given by Keyword Planner and the 12 month average we calculated based on the individual data points from the same source comes from using heavily rounded averages. Basically, Google Keyword Planner uses search volume buckets where you won’t get the true search volume, but whatever bucket is the nearest.

Russ Jones from MOZ noted that they found 85 different buckets for traffic, which are logarithmically proportioned. This discrepancy is miniscule when it comes to low search volume keywords, where the fragmentation between buckets has many small steps, like 10 searches/month and 20 searches/month. However, the higher you go in terms of search volume, the bigger the differences in bucket size. For example, the jump from 165,000 to the next bigger bucket of 201,000 gives room for a difference of up to 36,000.

The largest buckets show differences of up to 250,000 searches per month! It goes without saying that such a fragmentation makes it very hard to evaluate and compare the true value of individual keywords. But, it gets worse…

Google Keyword Planner’s bad data: combining search volume

Not only is the data obscured by search volume buckets, but different combinations of keywords have now been thrown together while only showing the collective search volume, without any indicator what keywords have been put together in the first place.

That’s what I noted before on Google Keyword Planner’s not giving any information about the keyword “designed”, and instead put it together with one of the others.

Here’s another example with “seo”, “search engine optimiZation” and “search engine optimiSation”:

They all show the same search volume, even though it makes absolutely zero sense. Just thinking about it logically, can it really be that the English version (with an S instead of a Z) has the same search volume in the US as the American version?

However, no need to speculate. Google Trends comes to our rescue!

As you can see, the difference in interest clearly shows that those terms could never have the exact same search volume. This isn’t new though, as Jennifer Slegg wrote about it on TheSEMPost.com some time ago. She also noted that Google Keyword Planner combines many search variations, including:

  • words that can be spelled with or without space (ie. car park and carpark)
  • words with and without punctuation (ie. kid toys and kid’s toys)
  • plurals with non-plurals for any word in the keyword phrase
  • acronyms with longhand version
  • stemming variants: -er, -ing, -ized, -ed etc keywords (ie. designer, designing, designed)

If you’re doing SEO using Google Keyword Planner or any tool that only uses its data without modifying it, then you should always check different variations in Google Trends to make sure you aren’t optimizing your pages and link anchors for the wrong variation of a keyword.

Also, it’s important to know how inflated the search volumes of those keywords become that way, so you don’t overestimate their value.

How to improve the search volume data

There are two solutions you can work out to improve the accuracy of your search volume data, one with a price tag and another one which requires a little bit of effort. First, those issues have been known to some SEO’s for a while now, so they tried to find appropriate solutions for that.

MOZ.com writes in their keyword explorer announcement that they modify their search volume data using Russ Jones' volume bucket methodology coupled with clickstream data from ~1 million US searchers. They also note that Russ apparently built a model that predicts the search volume range a keyword is likely to have with ~95% accuracy. It seems like a bold claim, and I can't really say if it's accurate or not. However, I believe that their data is certainly more accurate than if you were only to use Keyword Planner data. It is however questionable if it's worth the big price tag of a MOZ subscription.

Another solution would be Ahrefs’ Keyword Explorer 2.0, which basically does the same thing as MOZ in terms of modifying the search volume. They wrote in their Keyword Explorer 2.0 announcement that they also use clickstream data to modify search volume data and they found a way to ungroup the keywords, which I assume is similar to the MOZ approach.

However, comparing the data of those two, there are still major differences.

Keyword “SEO” from Ahrefs:

Keyword “SEO” from MOZ:

So, MOZ shows a lot less searches for “SEO” in the US than Ahrefs. This discrepancy increases even further with other terms, such as “search engine optimization”, as you can see here:

Keyword “search engine optimization” from Ahrefs:

Keyword “search engine optimization” from MOZ:

Apparently their methods aren’t as similar as first thought, but it’s certainly better than what we can get via Keyword Planner alone. While I own subscriptions for both services, if I had to choose I’d definitely take Ahrefs over MOZ due to the sheer amount of additional data those guys provide.

Most accurate search volume

However, there’s another, very expensive and old way to get accurate search volume. It has been used for a long time from professional SEOs, and you might already know about it. Simply create an Adwords campaign targeting exactly the keyword you want to know about. Let this campaign run for a specific amount of time (the longer, the more accurate your data), and use the impression data you get from Adwords as your search volume.

Of course, don’t forget to adjust it to the timeframe you’re looking for, usually one month. If you can make sure that your ad rank always stays high enough for your ad to show, then you should have the most accurate search volume available. Note though that using that technique for only a few days might give you bad data too, because the search volume of terms can sometimes vary greatly.

Last solution for the small budget

The last way to adjust the search volume data, for those with a small budget, would be to use Google Trends to put the keywords in relation to each other, to see how much they differ in interest. I already did that at the start when comparing the three terms “seo”, “search engine optimiZation” and “search engine optimiSation”.

While this is certainly the least accurate solution, it still provides valuable indicators of the differences.

Conclusion

Google Keyword Planner’s data has been shown to be heavily unreliable, even if you’re in possession of an Adwords account with high enough spending. The best solution in my opinion is to get an Ahrefs and MOZ subscription and compare the data of those two to get a better estimate of the true search volume. If your budget is high enough, or you intend to make a bigger project with heavy investment you can also opt for an Adwords campaign to further verify your data.

Alternatively, you can also opt for only one subscription, where my personal recommendation would be Ahrefs. Both Ahrefs and MOZ cost about the same with their monthly subscription, but Ahrefs is probably the biggest SEO data provider on the entire market and thus provides a lot more value for the same money.

Google Trends also might serve as a free solution to that problem. Either way, none of the solutions presented solve one issue you must deal with:

Finding and capitalizing on emerging trends in your market before your competition does. All solutions presented for the more accurate search volume use 12 month average, while TermChase.com is using hourly data over a week's period, therefore serves as the only solution to detect short-term trends early enough. In addition, it also offers lon-term trend data going back to 2003, which allows you to quickly identify keywords with long-lasting growth rates.

I’d highly recommend to try out TermChase.com with our 7 day trial, to see the value we provide first-hand.

Comments are closed.