How accurate is the Google Keywords Tool for SEO?

Being SEO & PPC obsessives, we spend A LOT of time using the Google Keyword Tool to predict traffic and therefore sales for our clients. The information presented by the tool help us decide which phrases to target and which ones to avoid. In fact, it's fair to say that the data from the Google Keyword tool actually helps shape our clients future online!

The information from the tool only started to show Google search volume in the last few years. But it seems that the data has not got any more accurate at all. Being in the unique position of overseeing many thousand page one rankings in Google, we thought we'd share our data with our readers to prove just how accurate the tool really is.

Some points to note on how we've presented the information.

• We've decided not to use the exact search phrases here to maintain client confidentiality so have instead just made reference to the industry & location that they operate in.
  
• In addition, we have also split the phrases up by what we deem to be general and specific. General refers to phrases which don't contain any geographical search terms or additional words / phrases which would reduce the amount of results returned considerably. Specific phrases are those which target a particular county or place or contain something specific to what the searcher is looking for. Ok, this isn't an exact science but it helps as a guide.
• The majority of our rankings use correct META tags and don't contain ...'s in them which might potentially influence whether the listings are clicked or not. Same for TITLE tags and page names as well
• We use a piece of software to record our rankings. This information is pulled from Google UK data centres and is accurate for the large majority of the time.
• Our traffic data is collected using standard installations of Google Analytics.
• All search engine listings are in what we'll call the traditional SERPs. Very few, if any, of the phrases selected have news, local,product or video results as part of the results returned, thus making it a fair test in that respect.
  

Here are the statistics collected;

(Click image for large version)

Ok, so there are a lot of numbers here. Some of the predictions from the Google Keyword Tool do not seem to correlate when compared to real clicks, while others are fairly good. Here are some of the findings that we can draw from the information that we've collected;

• Google Keyword Tool groups related phrases together when showing search volume. It isn't very clear how this information is grouped. For example, several phrases entered all returned the same search volume prediction, yet the specific phrases received very different amounts of clicks. Where the same search predictions were returned we omitted some of the possible results
  
• Overall, the amount of clicks compared to the estimated search volume from the Google Keyword Tool was just about 1%. When making predictions on what sort of traffic you can get from SEO, don't be too optimistic!
• There is a big difference between the clicks to prediction on specific phrases when compared to general phrases. In our test, general phrases get less than 1% of searchers clicks compared to something like 12.5% on specific phrases. So just because a phrase doesn't look like it will generate click volume, it may well actually perform better in terms of clicks than some phrases with supposedly higher search volume. This is another good reason to try your phrases out using a paid (PPC) search campaign.
• Unsurprisingly, phrases which rank at #1 get more clicks than those that don't. Our data shows that when a phrase gets to #1, it tends to get 2x more traffic than those that don't. This is a general rule of thumb and is obviously different depending on the phrases being searched for, the amount of traffic etc. It does however highlight the fact that your SEO efforts don't stop until you hit the top!
• A few of the phrases that we focus on had huge amounts of clicks compared to the amount of estimated searches carried out. These phrases tended to be very time specific and on newer phrases (e.g. containing 2009, 2010). This suggests that on phrases which are new or gaining search traction are not accurately estimated by the Google Keyword Tool.
  

All in all, it seems that the Google Keyword Tool can be helpful but in alot of cases, isn't very accurate at estimating clicks from search volume. Let me put that another way, using a straight line formula when estimating clicks from estimated searches isn't very accurate. Generally, it is a fairly safe bet to assume a higher click through rate on specific or local phrases compared to general ones. Yeah, so what, you probably knew that anyway. But maybe our research here might make you think twice about your estimates and possibly your ROI calculations.

Further research in conversion rates and PPC clicks to estimates would be good and if we have time we'll look into this further. For now, if you've any comments, we'd be very pleased to here them!