How to diagnose Google Penguin problems

August 24th, 2012 | Analysis 7 comments

It’s that dreaded scenario; you get up one morning and realize that something changed. Sales are down, traffic is dead, your website is having issues. But what is it? That’s often the hardest part.

Today we’re going to look at how one diagnoses a known problem such as the Google Penguin updates, from less established ones. The process is fairly similar for other algo updates such as Panda, so you should get a sense of how to diagnose those as well.

Establish if you’re a candidate

Before you get started, it might help to get a stronger understanding of the differences between a manual penalty at Google, and algorithmic ones. If you’ve have that, then go check over at Google Webmaster Tools. What we want to do here is look for any messages about manual action that has been taken on the site. If you have something like this, then it’s NOT going to be Panda/Penguin related entirely:

Google Unnatural Links Message

The reason we want to know that is to ensure we limit the options as to what might be happening. Sure, you may have been hit by both, but we’re getting ahead of ourselves. The main thing to understand is that manual actions are the ones you get notifications for, the Penguin updates are algorithmic and you won’t be notified. For the moment, let’s assume we haven’t gotten one of the GWT messages.

If you’re feeling lazy, here’s a 2 minute video to show the process (though I advise reading the entire article as well):

Connecting the dots

When one is looking to establish if there has been a Penguin issue, one of the first stops is the time line. We want to establish if it lines up with any of the known refresh dates.

Easiest place to start is on your analytic program of choice.

Isolate the time frame around the losses
Isolate Google (in GA; Traffic Sources > All traffic > Google )

One example looks like this:

Google Analytics Graph

This one can vary. In some cases it’s a smaller decline, other times it’s the ole ‘hockey stick’. This happens because sometimes only given terms have been affected, not the entire site.

Looking at Google referrer traffic we can certainly see:

Wednesday April 11 – 983 visits
Wednesday April 18 – 968 visits
Wednesday April 25 – 313 visits
Wednesday May 2 – 337 visits

In or about the 25th of April, Google traffic dropped some 2/3s in referrals. Once we have some dates, we can go to a list of known ones to see if there was any activity around it. I use this one over on the SEO Dojo.

Google Panda and Penguin updates list

Uh uh, certainly looks like we have something here. The initial Penguin release was April 24th 2012. Which most certainly coincides with the date of our losses. There were two Panda updates around it as well, April 19th and 27th, but we’re certainly closer with the initial Penguin release.

Instinct can play into things here as well. We can ask ourselves:

Do I have a lot of thin content or ads in contextual areas? (Panda)
Do I have a lot of low quality links? (Penguin)

Another place I like to check is Searchmetrics. They have what they call ‘Search Visibility’ which looks at rankings in aggregate. From this report we see:

Searchmetrics Chart

Once more, we have some further evidence that the site has indeed been whacked by the Google Penguin update. And in this case, seems there’s more to investigate than just the April hit, but that’s another story. Obliviously you can also then look at historic ranking metrics you may have on hand as well.

Where to next?

It will serve you well to avoid being myopic. We can never tell for certain if we’ve been whacked by the Penguin algo, that there may not be other elements in play. But in a case such as this one we’re looking at, I am certainly 90% certain that this is indeed a Penguin hit. So what should they do now?

There really isn’t a magic bullet for solving Penguin issues beyond getting rid of your crappy links and starting to build some quality ones to widen out the diversity. Where does one start?

As we touched on earlier, the first hint is often the size of the losses. If it’s a smaller drop, as opposed to a complete nuking, then we often find that certain terms have been hit, not the entire site. In those cases you want to try to establish which terms (core and modifiers) are the ones that were hit.

But we’ll get into that more next week. Stay tuned.

Dies ist nur ein Gravatar

David Harry is Sifu at the SEO Training Dojo and the president of Reliable SEO he works mostly in a consulting capacity to large corporate teams and agencies as well. He has been writing about the world of SEO (and information retrieval) since 2005 and also does podcasts and webinars on the topic.
Show all articles from .

Comments (3)

Comments (7)

  1. 2012/08/25

    Another way to check is Google Webmaster tools. Check your impressions. If those suddenly dropped after 4/25, that was most likely Penguin. People assume their rankings take a hit, but some sites just started showing up less in search for all their keywords, or at least for the money keywords that were in the anchor text of manually created links (link spam). If impressions drop, your traffic is sure to drop even if you’re rankings stay about the same. Again, you may not have lost your ranking position, it’s just for your money keyword, you’re showing up much less than you used to . . .

  2. 2012/08/27

    Thanks Chrisotpher for the add. Most certainly would be another good secondary metric to check as well.

  3. 2012/09/06

    The recent update is aimed at link diversity. The impact of Google’s changes will vary depending on how heavily a site’s link profile is skewed towards these factors. Some sites may lose rankings for everything, while others may lose rankings only on specific keywords.

Trackbacks (4)

Comments (7)

  1. 2012/08/24

    [...] Diagnosing a Penguin Hit, Searchmetrics [...]

  2. 2012/08/26

    [...] Diagnosing a Penguin Hit, Searchmetrics [...]

  3. 2012/09/01

    [...] a more detailed explanation head over to this post on Search Metrics where the diagnosis is [...]

  4. 2012/11/06

    [...] Harry did a blog post on diagnosing a penguin problem is an example of what you can do with a proper Universal Search [...]

Write a comment