Internet
How to
Newspapers
Twitter
SEO advice
Categories: How to, Internet. Tags: ,

Final proof that the Google Adwords keywords tool is useless in the UK

November 18, 2010 29 Comments

Google recently improved, supposedly, its Adwords Keywords tool - which is supposed to tell you search volumes for keywords, so you can plan your Adwords campaigns. Some SEOs use it to estimate natural search traffic as well.

I've finally proved that is of no use to anyone whatsoever in the UK - because it's showing data for UK-based searches using google.com, and includes no data for google.co.uk which is what we all use in the UK (it's fine if you're American and in the USA though). Thanks to Adam Sherk for running the numbers in the US for me. (Update Alternatively, it is showing UK data - but the global volumes are not based on google.com as Google says but on all versions of Google. See comments and follow up.)

Here's how you might use it. And why you shouldn't.

This is the data from the UK tool, showing exact match search volumes, with location set to all for the keyword "harpenden":

Searches for Harpenden, UK version of tool, locations set to all

Searches for Harpenden, UK version of tool, locations set to all

As you can see, the tool says there are 4,400 global searches a month. Because location is set to all, the local monthly searches shows 4,400 as well.

So let's assume you want to know how many people in the UK search for Harpenden - it's a small town in Hertfordshire, but you're not that interested in what anyone from abroad thinks of it. You might think you'd set location to the UK.

If you did, this is what you see:

Searches for Harpenden, UK version of tool, locations set to UK

Searches for Harpenden, UK version of tool, locations set to UK

Does the 2,900 in the "local monthly searches" column (now that location is set to UK) mean that there are 2,900 searches a month by people in the UK?

No.

Let's look at the US version of the tool. Here's what it shows when you search for harpenden with locations set to all:

Searches for Harpenden, USA version of tool, locations set to all

Searches for Harpenden, USA version of tool, locations set to all

As you can see, there are still 4,400 searches shown.

That's all the evidence I think we need that the Global monthly search volume is, as Google says in its help text, search data from searches on google.com.

What is the "local search" data then? Maybe Google mixes search volume data from .com with search volume data from google.co.uk when you set location to UK and look at local data?

It doesn't. I was almost 100% sure of this before, as it was impossible to ever get the local figure to be higher than the global figure - implying the local figure must be a subset of the global data for google.com (and not the sum of the UK searches on google.com plus searches on google.co.uk).

And here's the proof. This is the data for a more specific search for spire Harpenden:

  • In the UK, the global monthly search volume is 480/month.
  • This is what the American tool tells us it is as well.
  • The US tool also informs us that no one in the US searches for that term (local volumes there are 0).
  • The US and UK tools both say that, with location set to UK, local searches are 480 month.

Searches for spire Harpenden, US and UK versions of tool, locations set to UK

Searches for spire Harpenden, US and UK versions of tool, locations set to UK

This must mean that the local monthly search volume data is for searches in the UK using google.com. The tool can't be suddenly adding in .co.uk data when you choose local - or that 480 figure would have gone up (unless no one ever uses google.co.uk to search for spire harpenden. But that can't be true).

That's it really. When you're in the UK, the tool shows you searches from the UK using google.com. Most people in the UK use google.co.uk - so the data is completely meaningless, whether you're using it to run a PPC campaign or estimate natural search volumes.

Anyone care to disagree? (Previous posts on problems with the data include this one and this one).

You might also like
  1. … alternatively, the Adwords Keyword tool is spot on
  2. Why you shouldn’t use Google’s keyword tool for SEO
  3. What The X Factor tells us about Google’s keyword and search-volume analysis tools
  4. Google keyword tool: only certain ‘exact numbers’ show
  5. Proof and implications of javascript passing anchor text and page rank in Google

Share this post

Follow me on Facebook or Twitter

29 Comments »

  • Just one question, how have you excluded the possibility that the global volume aggregates from all of the local search engines (.ca .co.uk etc) and that the local volume is therefore the combined search volume from .com and .co.uk?

    • Good question. Fortunately, I looked into that!

      One reason is that the help text says global is for google.com. But that could be wrong.

      And the other is that I ran a load of foreign words through the tool and the search volumes looked very low - which said to me that it can't be aggregating data from all versions of Google ...

      • In my experience Google is pretty crap at figuring out what languages queries are in so the low volumes for foreign words might be a different issue.

        My own mini test:
        "carla bruni sarkozy enceinte" has 1600/month global on the all countries geo setting and French language setting. Has 1600/month global and local for French geo and French language. Has blanks for all the other geos I tried.

        Same results when I change from French to All Languages

  • Ok I just did a couple of quick tests of my own, search for something like warum, the German work for why, in all countries all languages, and you get 1,220,000 searches global and local, set it to searches from Germany in German and you get the same volume global and local, which makes sense because the majority of German speakers are in Germany. Then try searching for the same word, why. with setting to all countries all languages, you get 30,400,000 searches. If you set to US searches you get 16,600,000 local, UK gets 4,090,000 searches. Therefore the UK has a population 1/5 that of the US, but is accounting for 1/4 of the search volume for the word why. This would suggest to me that the volumes are taking into account all UK traffic, otherwise we would be accounting for a much lower amount of the search volume.

    • So do you think that the global search volume is based on all versions of google, and that the local one is based on the UK subset of that?
      That would also explain the data above - I must say I took as my starting point pretty much believing them that the global data is google.com data.
      EG in this post an official google employee said in September: "the updated version of the Keyword Tool provides search statistics based on Google.com traffic only"

  • I'm saying that I think, as @richardfergie said the data may be based purely on location regardless of TLD. The truth of the matter is I don't know any better than you, but the data you collected is open to multiple interpretation, and I don't think supports your ascertation that the keyword tool is no use for people in the UK. It's quite a large conclusion to leap to.

    • Logically, though, the data can't be based on location IF the global data is collected from .com only. As otherwise when you selected to see local data, the local figures should then exceed the global figures, as you'd have .com data plus .co.uk data added in.

      The figures above are also consistent with the global data being based on all (possibly english speaking?) variants of google, and then the local data really being based on UK-based searches across all variants of Google.

      That would mean the UK data was as you'd expect. But then the data would be problematic the other way - as it would mean that Google's stated source for global data (seaches on google.com) was incorrect.

      Either the global data is not as described, or the local data isn't local. Until we know, I don't see how anyone can rely on this data to do anything.

    • Pokie guy says:

      I agree with you. Global is all the country versions + .com and local is the local version and maybe .com searches also included - we do not know as google may give aggregated data from both local + .com based on ip. So if I am in the UK and use .com to find something this data is added to local even if I use .com. Why? Potential advertisers market is the answer - advertisers need data on the volume and it is for people living in that area - whether they use .com or the local co.uk search engine. Google does not say that local searches are based on local versions of google engine only! You will never know search volume of co.uk but you will have data on searches conducted locally on both .com and co.uk - and will see global searches data which contain local (with the .com) and other .com (as searches made in France for the same using .com is not added to local at all.)

  • "As otherwise when you selected to see local data, the local figures should then exceed the global figures, as you'd have .com data plus .co.uk data added in."

    What I mean is that the starting point for global is supposedly .com data. So when you ask about local, if it adds .co.uk searches into the local figure, that number should then be bigger than the initial global figure. This never happens though.

  • jaamit says:

    I dont think this is true Malcolm.

    OK another test on a term I'd expect to have relatively higher volumes in the UK vs worldwide or the US - "arsenal".

    here are the figures the tool comes out with

    UK: http://imgur.com/Lv1gg
    USA: http://imgur.com/bl0il
    All countries: http://imgur.com/YtmU9

    UK searches on the exact match term are 10x bigger than US, and all countries is about 2x bigger than UK alone.

    Broad match pattern is similar, but less pronounced with UK 2x bigger than US.

    It doesnt feel like this could be on the basis of UK visitors on google.com alone. I think the quote above from a google employee was in relation to being Google alone vs search networks - so he used Google.com for clarity - dont think he thought about the other ccTLDs.

    But we need a clarification on this particular question from a Google employee I guess...

    • The help text on the tool itself says global data is: "The approximate 12-month average of user queries for the keyword on Google.com and the Google Search Network".

      How do they define the Google Search Network? Here they say "The Google Search Network: Ads are targeted based on a user's search terms. For example, if you search for "Italian coffee" on a search engine powered by Google, such as AOL.com, you'll see related coffee ads next to the search results." which makes it sound like it's all the ads they run on non google sites.

      Though that quote earlier said they no longer count "traffic from search partners" - is that the search network and therefore the help text is wrong?

      If they can't tell us what the data is, how can anyone expect to use it?

  • I did a follow up post saying that it is possible that the UK data is accurate if the global data is based on all versions of google, and not just google.com.
    If that is the case, their help text is all wrong ...

  • Brennan says:

    It seems overall that the keyword tool for adwords externally isn't very accurate. The best way to get a good reading is to actually go with the internal tool for adwords just like you were bidding on the keyword. Much of the time is shows a different number and can give you more data.

  • Pete Lovick says:

    Guys, it seems to me that there's some simple confusion going on here.

    When the official google employee said in September: “the updated version of the Keyword Tool provides search statistics based on Google.com traffic only” ..he simply meant that they are no longer including results from other search engine networks like Yahoo and Bing. They are now only going to show results for the whole of their own google search engine network only, (which includes results obtained via all their agreed licensed users of google search, like for example, AOL).

    In their Help file, when they refer to other search networks, they are referring to other engines or companies, which they previously used to amalgamate the results of in with their own results until they just changed the format a few months ago to only provide their own. (..probably because they are now the most used search engine company so why should they include other engine's results too? ..who knows?)

    Also, whether we as users in the UK access google's search network results via their .com or their .co.uk TLD is totally irrelevant, as we are still accessing the results off their whole worldwide database. ...which is why any Local results it spits up is the same via either TLD because it just knows where you are accessing their database from via your IP address. (i.e it doesn't care whether anyone has accessed their database via a .com or a .co.uk TLD. The results it gives you will be the same).

    Similarly, for example, anybody in France who accesses their search database via a .com or .fr domain doesn't matter to google. It'll still direct them to their database and tell them how many searches from people with an IP address in France searched for the term requested. i.e. It doesn't matter, in any country, which TLD anyone uses to access google's database, as it'll only spit out results from people with an IP address in that country that have previously searched google's database for any given term.

    Knowing this, now you can see why no country's local search results could ever possibly be higher than the total global monthly searches throughout the whole of it's network. By Local it means people with an IP address in that country only that have searched for a given term ..and not people who accessed their database via any countries typical TLD.

    In a nutshell, google's .com domain, or any of subset country TLD's, are not connected to google's actual database results it gives out. Only the IP address of any google searcher is what matters, and this is the results it's giving back, for example, how many UK IP addresses have searched for a term.

    Avoid connecting any country's TLD with results and you'll see it's giving you the results of people with that country's IP addresses that searched for a term - and not which domain/TLD they used to access google's database.

    Apologies for this being a bit longwinded, but it's the only way I could hope to make this clearer. ...unless I've got it all totally wrong of course. lol. ;-)

    • Hi Malcolm

      It looks like you have been misinterpreting the Google keyword tool results in this and previous posts of yours about the Google keyword tool.

      Pete Lovick has explained it correctly.

      Folks - the local results are based on IP address of the searcher.

      Don't expect the results to be accurate - they are a sample of all searches, and are statistically valid when compared to other data provided by Google that is gathered in the same way, if the sample is large enough. Don't trust small numbers for obscure searches. Watch the trends, and look at relative search volumes rather than exact numbers.

      Do some real tests - not just looking at what the keyword tool tells you for different countries, but what your Adwords ads or organic search results get in terms of clicks. Once you have a reasonable amount of data to look at ("reasonable" will depend on your organic ranking or adwords spend, and actual search volumes) you'll be able to estimate what ranking for a particular search term will mean in terms of visitors to your website. After all, that is what you are interested in doing if you are using the keyword tool. It is a tool, and should be used for the job it was designed to do.

      There is no "perfect" keyword tool. All are subject to some bias, and looking at multiple tools will probably just be more confusing than informative. I've found the Google keyword tool to provide reliable data - it is based on the largest sample of actual internet searches as Google has the lions share of the search market.

  • Rohit Kanwara says:

    Hi Pete,

    I have been following this blog for quite sometime. I am currently using google keywords tool to try and identify right keywords for my website and to be honest i am quite confused whether it is working correctly or not. I have been using the keywords tool for about 2 months now. But the problem is that I don't get enough traffic coming to my website. For eg: for one of the keywords, the tool suggests that there are 4,400 local (UK) searches for it in a month. When I type the keyword on google, my site comes up on the first page renaked 4th on the organic search. Still i only get 50-60 people coming to my website in a month.
    I am trying to find the best possible fit for it, but right now i am just confused. Will appreciate if some sort of advise on the google tools or any other tool could be shared.

    thanks

  • Pete Lovick says:

    Hi Rahit,

    There's only two things here I can suggest:

    1. When using the Google Keyword Adwords Tool you should only ever take the results that come up when you are logged OUT of your google adwords account, if you have one. i.e. Don't 'login to get the full list of results' or whatever it advises, as these results are skewed with google's adwords PPC performance data.

    Select UK at the top as normal and make a note of the keywords that come up when you are logged out.

    2. You also need to be aware of the organic search traffic you get from the individual positions on google's page one.

    Currently the top 3 take the lion's share (...from memory about 45%, 18% and 15% respectively about 6 months ago for the 1st, 2nd and 3rd), positions 4-9 get about 1-2% of the traffic that hits the page and position 10 gets about 4% (because some people scroll to the bottom) and the left overs are equally split with the PPC paid ads down the right hand side.

    1-2% of 4400 UK monthly results IS about the 50-60 visitors per month you are currently getting, so ease up on yourself because you are in fact doing it correctly. No panic!

    I don't want to say anymore here because I'd be getting completely off Malcolm's topic here in this thread, but basically you want to check your keywords as in Item 1 above, and if they are correct then relax, as pointed out in Item 2 above you are getting the right amount of traffic.

    Your aim should then be to a) move up a few positions to get more traffic, and/or b) select alternative keywords you can spread throughout your internal site pages then repeat your page ranking process to get your internal pages up to position 4 or even higher as well. Go for UK monthly searches of between 800-3,000 as they'll be pretty easy to rank for and will dramatically boost your free traffic to your main keyword index page if your Home page navigation on your internal pages is clear to your visitors.

    Hope this helps.

    (Great job btw. ...position 4 is NOT to be sniffed at!).

  • Emma says:

    can i ask what you guys would recommend for keyword research/discovery instead??

    thanks for the great post :)

  • Tomasz says:

    You seem to be bent on proving that GKT is useless, while in fact after reading this article it seems to me that you just don't know how to use it and how to interpret the results. Sorry.

    • Tomasz: Well, to be fair I do link to a follow up post in which I say maybe I'm talking rubbish. So I'm not sure I'm bent on proving anything.

      Having said that, it's definitely not accurate....

      If you've ever used it to estimate traffic and then got a number one ranking, you'd be amazed at how little the traffic you get resembles the figures you get from the tool.

      Bt it if works for you, then good luck with it.

      • Jonathan says:

        Have you tried comparing data from keyword tool to webmaster data for a site with very strong first place position for that keyword and checking impressions, or comparing a high budget Adwords account impressions for that keyword to see if they are close?

  • Anders says:

    Hi Malcom,

    With all the questions raised here about the validity of your conclusion I would recommend that you change the heading and text of the post.

    As it stands now it is highly misleading, and there is enough disinformation on the net about this stuff as it is...

    Just my two cents...

    • Anders: The data is misleading. Don't use it. That's my view. You can read more in the comments over at http://www.malcolmcoles.co.uk/blog/why-you-shouldnt-use-googles-keyword-tool-for-seo/.

      • Anders says:

        There is a material difference between saying "My view is that the data is misleading and I do not think you should use it" and "Final proof that the Google Adwords keywords tool is useless in the UK".

        If you know how to use the data and what you can use it for you can gain very valuable insights - that is my experience and opinion.

        However you are knowingly publishing wrong information here ("Well, to be fair I do link to a follow up post in which I say maybe I'm talking rubbish.").

        The Google Adwords keywords tool is NOT useless in the UK!

        As with any other tools and data you need to know how to read it and how to use it.

        You should take this post down or change it... the same goes for the other post you are referring to.

  • Rick says:

    Malcom -

    I've been professing the gross inaccuracies of Google's search estimator tools for years. I have never found a satisfactory answer from Google, and as illustrated in this post, I get just as much push back from the general marketing community.

    It would be one thing if the estimates were just wrong (which they are - I've run hundreds of test to measure their accuracy), but it's doubly frustrating when you come across logical contradictions like those you've addressed that don't seem to ever be resolved.

    If Google wanted to clear this up it would be this simple: provide documentation of how the numbers are arrived at outside of the vagueness of two sentences.

    Anyway - I completely agree with you. Thanks for the post.

Leave a comment!

Add your comment below, or trackback from your own site. You can also subscribe to these comments via RSS.

Be nice. Keep it clean. Stay on topic. No spam.

You can use these tags:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

This is a Gravatar-enabled weblog. To get your own globally-recognized-avatar, please register at Gravatar.