SEMrush studied 830 backlink profiles of websites hit by link penalties.
What they discovered are insights that may influence how you think about link building.
Not a Ranking Correlation Study
Ranking correlation studies try to understand what the ranking factors are by studying millions of Google search results pages (SERPs).
Those kinds of studies lead to unreliable results that should not be taken seriously.
This research by SEMrush is not a ranking correlation study. Which is why I am writing about it.
A study like this shows some interesting insights.
However, because Google processes are not openly discussed by Google, one can’t say with 100% certainty that a certain pattern led to a link penalty.
But because this isn’t a ranking correlation study, what is seen can still be remarked upon and debated.
1. Guest Post Penalty Patterns
The SEMrush study found that many sites that suffered manual actions for guest posts had specific patterns.
SEMrush observed that sites that had certain link features together tended to be the ones suffering from a link penalty.
A notable pattern was guest articles that were clearly labeled as sponsored plus an aggressive keyword-rich anchor text on followed links (i.e., links that aren’t modified by a nofollow HTML link attribute).
In my observations of sites that have received link penalties, aggressively adding keyword rich anchor text to links tended to be a factor in the link penalties.
Whether the label indicating that a guest article was “sponsored” played a role is debatable.
But I believe that most search marketers with experience can agree on is that aggressive anchor text optimization can be trouble.
Maybe it’s possible that the two factors together, the sponsored label and the keyword optimized anchor text makes red flagging these links easier for Google.
According to SEMrush, common phrases in the sponsor labels were:
“The article was sponsored by
Information was provided by
Advertorial post on behalf”
It Takes More than One Bad Link”
Another interesting pattern SEMrush discovered was that it takes more than just one links to trigger a penalty.
Websites that received penalties tended to have patterns in the links that seemed to indicate someone was creating the same kind of link over and over, SEMrush found.
According to SEMrush:
“Based on our research, we discovered that a single bad link shouldn’t cause a lot of trouble. A penalty usually ensues once Google detects a link manipulation pattern. So, you have to show a combination of 2-3 or more unnatural links to get penalized by the search engine.”
Personally, I think it takes far more than two or three unnatural links to get a penalty.
2. Doesn’t Matter How Old Links Are
How old the bad links are don’t matter, according to the study.
This is something I’ve seen before in the first wave of Penguin penalties many years ago.
Guest article links were being called into question by Google and a pattern I noticed was that all of those sites had paid links that they had purchased years before and had forgotten about.
This is SEMrush’s observation:
“Google makes no distinction between recent and old unnatural links. So, if your backlink profile wasn’t impeccable a few years back, your current spotless link-building efforts could be washed down the drain.”
That observation matches my experience with clients that came to me with link penalties.
If a site has ever engaged in paid links or questionable link building, it may be prudent to file a disavow now in order to avoid having those old legacy links contribute to a penalty later on.
3. Unnatural Links Can Happen on Any Site
This finding is interesting and I kind of agree with it to a point.
SEMrush said that it doesn’t make a difference if a spam link came from a low quality site or what is generally perceived as a “trusted” media site.
The way I’ve always built links is that no site is trusted.
Every site is as good as it’s outlinks.
I know that certain so-called “trusted” media sites sell links.
Even without knowing this as a fact I can know this simply be examining where they’re linking to and how they’re linking.
If I can figure this out in less than a minute then I have always figured that Google can figure it out and stop those links from counting.
I call those kinds of sites “poisoned.”
For example if the site is about some particularly industry like hotels or home improvement but it contains articles about link building or SEO, for link building purposes I’ve considered those sites poisoned and wouldn’t take a free link from a site like that.
So maybe it depends on how one defines “trusted” because what generally passes as “trusted” and “authority” in my opinion needs closer scrutiny.
Here’s what SEMrush said:
“Our study only confirmed that Google is an equal opportunity search engine. It makes no difference to Google where the links to your site come from — be it spammy, dubious websites, or trusted media.”
See what I mean? Don’t assume that just because it’s a big media site that the link is not going to be harmful.
The standards for judging a site must be more realistic than the Moz DA score of the home page or the quarterly earnings of the media site.
That’s not a knock on Moz. Their DA score is engineered to be a comparison tool, not as a link building yardstick.
Even Moz will tell you that it’s a misuse of their tool to use it for making a business decision like link building. It’s not designed for that use.
4. Google Penalizes the Entire Site
SEMrush observed that when Google drops a manual action penalty on a site the rankings of the entire site tends to suffer. This contradicts the view that Penguin is granular and only affects certain pages.
“Although Google’s Penguin Update implies a more granular approach when it comes to penalizing sites for spammy links, our study counters this notion. In fact, we discovered that Google is more likely to penalize the entire website rather than stick to its granular approach and just hit the specific URL/site section.”
5. 50% of Penalties Involved Anchor Text Abuse?
SEMrush observed that half of the manual actions involved aggressive anchor text optimization.
That means using the keywords that you want to rank for as the words used in the link (the anchor text).
We cannot know for certain if overly optimized anchor text is the explanation for 50% of the link penalties.
Google can communicate that information but Google’s spam team tends to be general and opaque.
They send examples but don’t always specifically say it’s the anchor text.
But… that tends to be a feature of a lot of link penalty cases I’ve looked at.
Here’s their finding:
“Anchor text can really make an impact. We discovered that half of the penalty cases involved links that use anchor texts that exactly match a keyword the website wants to rank for in organic search.”
6. Google Is Vague About Link Penalties
This matches my observation above about how Google is vague about what exactly caused the penalty. Google sends examples of what caused the penalty. It’s up to the publisher to perceive what is wrong with the link.
This is unfortunate because a typical publisher might need to have this spelled out.
What’s obvious to an experienced SEO is not apparent even to an SEO with a few years of experience, much less to a publisher or an eCommerce site owner.
This is what SEMrush observed:
“If you expected that once a penalty is issued, Google will pinpoint which links they considered unnatural, you are wrong.”
7. Link Penalty Removal Takes a Long Time
In their study of over 800 sites, SEMrush observed that it can take a long time to get a link penalty removed.
In my experience and my understanding, Google wants to see effort by the publisher to remove the links. They want to see evidence that the publisher understands what caused the penalty and to state that they won’t do it again.
It’s generally not enough to submit a disavow file and wait for good news and sunshine from Google’s spam team.
Even in the early days of SEO when I could just send an email to someone at Google or Yahoo on behalf of a client, the client would still have to be squeaky clean over and above the normal state of a site.
So if a site had a link problem they had to make sure that everything was taken care of to a level that could appear to be way beyond what caused the penalty in the first place.
What SEMrush found seems plausible:
“Once you fix your toxic links, you may need to submit 2-3 reconsideration requests before the penalty is lifted. It doesn’t sound so bad, but the whole process may take up to 6 months.”
8. Link Penalty Avoidance
They say that the best defense is a good offense. What that means for SEO is that it’s a good idea to be proactive about links.
If someone in the past used poor judgment in their link building, it’s best to take action now than wait until it’s too late.
The best approach is to make those links go away.
Failing that, the last recourse can be a disavow file.
But the best approach should be to have those links disappear for all time so that you don’t have to file a disavow.
Read the SEMrush study here:
Google Penalties Research: Detecting and Dealing with Unnatural Inbound Links