Google EMD update and PMD decline over time
By Andy, Oct 22nd, 2012
When Matt Cutts or the Google Webmaster team release information on an algorithm change, social networks, forums and comments boxes are all at a flutter with disgruntled webmasters claiming the big G is out to get them. It’s nothing personal, Google has its own agenda and I’m sure Mr Cutts hasn’t got a big dart board with your face on it. One thing I do when I receive a piece of information like this (the algorithm change not that Mr Cutts has my face on a dartboard) is check out the data myself, rather than taking Google’s word for it.
Exact Match Domain (EMD) Algo update
With this in mind I wanted to review the recent EMD (exact match domain) update, which looks at lowering domains of low quality that match the search query. Changes have actually been in place for a long time following a change by Google allowing these domains to rank in the first place! Take a look at our findings for both ‘exact match domains’ and ‘partial match domains’.
Many thanks to our junior search engineer Danny Wood for providing data and insight to this post.
Here is the tweet that was picked up by many SEO news blogs:
New exact-match domain (EMD) algo affects 0.6% of English-US queries to a noticeable degree. Unrelated to Panda/Penguin.
— Matt Cutts (@mattcutts) September28, 2012
Great change wouldn’t you say? Why should a domain rank simply because it is an exact match of the query? Surely there should be multiple factors that influence the top ranking of a website, no? Google will no doubt suggest there are hundreds, but from what I saw in 2010 – EMD/PMD ran riot.
I’d also like to make the point that hopefully you find this blog interesting and useful. My style is to look at data, put across a couple of arguments and then welcome a discussion. You won’t find any cringe worthy names for algorithm changes nor will you find me wearing a novelty Google T-Shirt whilst tweeting Matt Cutts for approval. No. The argument may conflict with what US SEOs have experienced, considering changes are rolled out at different periods (we didn’t get Panda 1.0 until April 12th 2011!).
We have collected data for EMD and PMD across a wide search terms that fall under a number of different categories. The maximum number of words used for a keyword phrase is three, whilst the domain target range is ‘word1word2word3’ or ‘word1-word2-word3’.
Below is a collection of this data:
Table showing number of exact match domains (EMD) and partial match domains (PMD) in Google’s 100 search results, over a three year period for a wide variety of keyword terms.
The table shows the amount of exact match domains and partial match domains within the top 100 of Google.com for a wide range of keyword search terms on the 1st September over a three year period and one the 1st October 2012, a few days after the news was announced. It is clear to see that 2010 and 2011 had a significant amount of EMD/PMD inclusions. Below is a chart highlighting he movement over this time for EMD/PMD for ‘watch movies’:
Chart showing increase in amount of EMD and PMD in 2010 and 2011 before reducing in 2012
This is interesting as there is zero movement from 1st September 2012 to 1st October 2012. The biggest shift comes from September 2011 to September 2012
An EMD that was hit
The following chart highlights a testing domain that I have, which was impacted by the change:
An example of a three word exact match domain that was ranking on page one in Google.com that dropped outside the top 100 when the EMD update was rolled out.
This domain is a three word domain (www.word1-word2-word3.com) and is (in my opinion) a relevant result to return for the query in question. However, this is of course subjective and personal opinions shouldn’t come into what is actually happening right now. Here is a little bit of information on the domain and some reasons as to why I think it was hit:
- The homepage content has not been updated for some time
- The information is ‘tournament’ based, so isn’t always relevant to current events
- Limited number of backlinks over past two years
- Redundant news/blog section that hasn’t been updated since December 2011
- There are no advertisements on the website
I believe that a combination of these issues and the fact the domain is a three word EMD is the main reason as to why it has dropped outside the top 100. This is a great example of a EMD being hit the day Google announced that this type of domain was being targeted, however – I think that a lot of work had gone in much earlier than the end of September.
The following chart highlights the movement of five EMD over the course of a three year period, but notice how three of the domains are limited to the launch of the Mayday update and the end of April 2012, when I noticed a massive drop in EMD/PMD inclusion:
Chart showing EMD for the keyword ‘Golf’ over the past three years and the decline of .fi, .co.uk and .com.au within Google.com
So the above chart shows EMDs being hit way before Matt Cutts tweeted. This backs up my theory that this has been in motion for some time and September 28th 2012 was just another release of the EMD update. Also, if you’re new to SEO, have an EMD and are worried – then look at golf.com and golf.co.nz – both did not drop in positioning following the latest release. It has been mentioned that, like Panda, this is released periodically – so, I’m sure there will be another mention in a few months time.
We’ve covered EMDs, but what I find more interesting is the impact PMD have within the SERPs. I was quite vocal about the state of the UK SERPs from June 2009 (when there were a number of foreign non-related websites shooting up the rankings) to September December 2010 – which in the summer of 2010 saw a ridiculous amount of ‘keyword rich’ domains with limited history and a poor backlink profile ranking on page one.
Let take a look at the first 13 EMD/PMD in the top 50 for ‘Gucci’ on 1st September 2010 and see how they rose in positioning and subsequently, declined:
Charting showing top 13 PMD over a three year period within Google.com.
Interesting that all 13 domains used within this chart didn’t rank in the top 100 come 1st June of this year. What is also interesting is the influx of domains around the summer of 2010. I remember seeing this influx when analysing the movie sector (following the May Day update) and complained to Google suggesting that newly created ‘keyword rich’ domains were ranking too high for the quality of their service. Yes, it may all be subjective – but there wasn’t even content on some of the websites!
If we filter the results to those that rose over the summer of 2010 (using the same key as above) you’ll notice a sharp rise and then decline, that coincides with Matt Cutts suggesting at Pubcon that ‘exact domain matches are next‘ on the agenda:
Chart showing surge in PMD over the summer of 2010 for ‘Gucci’ keyword.
The Gucci case even highlights the keyword match domains ranking before the May day update:
Chart PMD filter being turned “on” and “off” prior to May Day update.
Again, extremely interesting that all three of the above ‘rose’ and ‘declined’ at the same time. Perhaps testing a filter of some sorts? If we take a look at one of the domains (www.honeygucci.com). which now redirects to the following URL and page:
Should a domain rank simply because the keyword is in it? Clearly not. Anybody would probably agree with that, but if the brand offers a superb service and is well known (take B&Q’s used of www.diy.com for example) then it should be a top ranking domain for the keyword query.
However, I believe – from my experience in working closely with EMD SERPs – that the real problem happened over two years ago and only in the last year has Google decided to tackle this, which is why you see a massive drop from September 2011 to September 2012 for EMD and PMD. This algorithm “filter” was rolled out in the past and April 2012 saw a big drop (see Golf chart) for EMDs and PMDs within the SERPs. I reviewed a number of websites and I understand the reasons as to why they have been weakened within the SERPs. It appeared that some were only ranking due to the keyword within the domain!
To conclude – this is a welcome change but has been active for quite some time. If you have any questions on the above or would like me to look at a particular keyword for you, then please leave a comment.Share
Sometimes you catch glimpse of one too many similar...
Yesterday, in another eagerly awaited video from Google’s Head...