7. Site Autopsy – A look at the backlinks 2


In this post, I want to have a preliminary look at the backlinks to the penalized site.

In the previous posts in this website autopsy series, we looked at on-page factors, and tried to find ways in which the site could be improved.

If you have been following this series, you’ll know that the site was originally hit by Penguin, and that the website owner did not think Panda had caused problems.   If Penguin was the culprit, then any penalty may well have been down to backlinks.  This is where we have a bit of a problem.  The penalty was a long time ago, and the website owner has taken action on his backlink profile since the penalty.  What we see today was not the profile at the time the site was penalized.  Therefore it is not possible to find specific backlink problems that existed at the time the site was hit initially.  However, having said that, the site owner thinks the site is still penalized, so it is possible backlink problems remain.  That is what I want to have a look at now.

This will be a preliminary look at the backlinks, and if I think there is a need for a more thorough investigation, I will do that in a later post.

To look at the backlinks, there are a number of tools we can use.  The one I have used the most over the years is SEO Spyglass.  However, for this initial investigation, I want to use an online service called Ahrefs.  This is a subscription based service, but I am going to use the free trial version to look at the backlinks to the Cayenne pepper site.

Over on the Ahrefs site, you can try it for free with a URL:

image

You’ll get back limited backlink data for the URL, but it is still very useful for an initial look:

image

That top line is total referring domains (right side of y-axis shows values), and the bottom line is total referring pages (left side of the y-axis shows values).

I have set this graph to show data going back as far as possible (by selecting “ALL· in the tabs at the top).  We have data going back to March 2013.  What we can clearly see is a rapid drop off in total referring domains from March 2013 (around 500 domains linked to the site) to September 2013 (about 360 domains linked to the site).  This does recover a bit into November 2013, but then falls off again to the present day.

If we look at the number of referring pages, we can see that the drop off in numbers is initially slower, but over the total timespan shown, Ahrefs is reporting that total links to the site went from over 3000 down to around 300.  This shows that a lot of links were lost during this time.  That would be a major cause of ranking losses.  This is shown dramatically in another graph that ahrefs offer us:

image

On the top of the x-axis is the new links that are acquired.  Ignore whether they are dofollow or nofollow.  Below the x-axis you can see the links lost over time.  Since the start of this time-frame, you can clearly see the volume of links lost.

Why links lost?

There are a number of reasons why links are lost in volume like this:

  1. During the introduction of Panda and Penguin, link networks that were used for backlinking closed, and lots of pages were removed from Google.
  2. A lot of sites removed backlinks from user-submitted content to try to comply with Google.  This included mass deletion of pages that people had submitted.
  3. After the Penguin penalty, the Cayenne pepper webmaster actively tried to clean up his link profile.

So, even if this cayenne pepper site fully recovered from any penalty that it might have been given, it’s rankings would probably not recover fully, since before the penalty, all of those backlinks were contributing to the rankings.  After any penalty is lifted, there is a lot fewer links to help rank the pages.

Number of Links per Domain

I would like to draw your attention to the ratio of domains to links. 

In March 2013, 500 domains supplied 3000 links to the site.  Therefore on average, each site supplied 6 links back to the target site.

If we look later on in September2013, around 360 domains supplied 2500 links, or around 7 links per domain.

Fast forward to May 2014, and 349 domains supply 1146 links, or 3.28 links per domain.

I usually find that a high number of links per domain is a bad sign.  Things improved with this, going from 7 per domain down to 3 per domain, but I expect there are still problems there.

So what is happening? 

Well, without knowing more about the links back in early 2013, it’s impossible to say, but we can guess.

If a site is sending a lot of links to another site, there could be a number of reasons.  For example:

  1. The site in question is using a script that has a problem with URLs, and essentially uses a number of different URLs that resolve to the same web page.   Over the years, I have seen this happen with WordPress comments on some sites, and with some forums.  What is essentially one link on a site, is seen by the search engines as dozens of different links so these tools report them as multiple links.
  2. The site has sitewide links to the target site.
  3. The site is controlled by the same webmaster as the target site, and they have used it to feed links to the target site.
  4. The site is some kind of submission site and the webmaster has posted multiple submissions to the site.
  5. Any other reason you can think of.

Generally speaking, from my own experience, when a site links out multiple times to the same domain, those links are lower quality, and often the site that is linking to the target is lower quality.  I find that this scenario usually indicates poor backlinks in the profile.

Generally, in my opinion, it is more unusual to have multiple QUALITY links from a single domain, though there are of course exceptions.

Does ahrefs give us any information that might help?

Well it does tell us that there are 5 sitewide links:

image

Sitewide links are links that appear on all of the pages of another website, and they are a bad thing.  I can look at these in ahrefs:

image

This is a little confusing to me.  Ahrefs shows me 5 links, but it really is only 2 websites.

The top one is campesinossosa.blogspot.com

It has the following in the sidebar:

image

Checking Google, this site has 428 pages indexed.  Assuming this link appears on all of the pages, that’s 428 links back from this site.  It is a low quality site and these links should be removed or disavowed.

The second site is another blogspot site go-nine.blogspot.com.

This one has the following in the sidebar:

image

Looking at Google, this site has 552 pages indexed.  Again, it is a low quality site, and these sitewide links should be removed or disavowed.

The fact these are both Blogspot sites may indicate that the webmaster of the Cayenne site owns these two sites, and setup some kind of auto-posting to them to build them out.  However, I think it is more likely that the Cayenne site owner does not own these sites, since these sites link out to a large number of other websites as well.  I would guess that either the Cayenne owner asked (and possibly paid) for these sitewide links.  Even if this is not the case, these links are toxic and should be removed or disavowed.

Looking at the fall-off in inbound links (and domains linking) to the cayenne site, I would suspect that there were a lot more of these sitewide links initially.

Image backlinks?

In the “Backlink Types” screenshot above, I can see there are 14 images used as backlinks.  These are something that always raise a red flag when I am looking at a penalized website.  Why would a site link to another site using an image?

Well, there are legitimate reasons for this, for example, there are sites that show a thumbnail of the site to link to it.  These include directories, and website stats sites.  While these two examples are not something to worry about, I would look at where these images links are coming from, and make sure they are not just images placed on a page for a backlink.  Any image that links to another site without very good reason, should be removed or disavowed.

Anchor text used in backlinks

Anchor text is a difficult one.  Links should be natural, so that means anchor text should be natural too, and knowing what a natural anchor text is for a site is just about impossible to guess.  One thing I would say is that the site name, in this case “Cayenne Pepper” SHOULD be one of the higher percentages, simply because that is what the site is called.  Where I would be concerned is when commercially valuable keyword phrases start appearing in any volume.  Here is an anchor “cloud” showing the most used anchor texts:

image

Without full access to ahrefs, it’s really not possible to look in too much detail at the anchor texts used on external domains.  I may revisit this in a later post.


Leave a comment

Your email address will not be published. Required fields are marked *

2 thoughts on “7. Site Autopsy – A look at the backlinks

  • Ian Smith

    Andy,

    Thank you once again for your terrific insights.
    As you may remember I have been struggling to be an overnight success on the internet for around 20 years and have often been in despair as my websites once again are dumped by Google, and I am certain I am not alone.

    This time I have stopped trying to create backlinks, I have stopped all pretensions at SEO, and have ignored Adsense. I have just tried to write content that I’d like to read, and only one website.

    The only ‘marketing’ I am doing is through Twitter and I try to follow 30 people a day, but I just checked my backlinks and I seem to have over 180!!

    For months now Piwik has shown between 0-5 visitors a day, most of them me!!

    In the past two weeks my visitors are climbing steadily, my return visitors are climbing slowly, my Bounces are down to the middle 50%, and I am now seeing 25 visitors a day. I am also getting a number of retweets and favoureds on Twitter.

    I have used curation, and also because it is a travel site, hotels and restaurants are part of the articles, I am finding that by writing to the owner of an establishment asking if my entry is OK, I can see visitors from this source.

    My articles are also a lot longer, no more the 300-450 words, my recent article was 4000 words

    In the long run is this method viable, because it seems to be working albeit slowly for now.

    Thank you again for your inspiration

  • Steve V

    Andy, thanks for posting this article. This gave me some really good techniques to troubleshoot linking issues that I can use for my sites!