Ok, maybe Google doesn’t find your site loathsome. But, Google definitely finds your site unnatural. Specifically, they find your link profile unnatural.
While this isn’t a brand new effort (Barry Schwartz reported on Google’s unnatural link profile messages back in Jan of 2011), Google seems to have kicked this program into high gear in the last month. Recently, several people have received this message in their Google Webmaster Tools inbox with the title: Notice of Detected Unnatural Links. Since one of these has already been published in Google Webmaster Help Forums, I’ll just use that one so you can see exactly what Google has to say:
Dear site owner or webmaster of http://xxxxxxxx.xxx/ We’ve detected that some of your site’s pages may be using techniques that are outside Google’s Webmaster Guidelines.
Specifically, look for possibly artificial or unnatural links pointing to your site that could be intended to manipulate PageRank. Examples of unnatural linking could include buying links to pass PageRank or participating in link schemes[emphasis mine].
We encourage you to make changes to your site so that it meets our quality guidelines. Once you’ve made these changes, please submit your site for reconsideration in Google’s search results. If you find unnatural links to your site that you are unable to control or remove, please provide the details in your reconsideration request.
If you have any questions about how to resolve this issue, please see our Webmaster Help Forum for support. Sincerely, Google Search Quality Team.
What’s This Letter Really Mean?
A couple of things. Primarily, your site’s been penalized, as evidenced by the reconsideration request. Now whether your website has been completely removed from the index is suspect, but it’s safe to say Google has banished it to the nether regions of the index; nevermore to see the light of user’s queries. Secondly, and perhaps, more importantly, it is evidence that the algorithm can detect link manipulation (or at the very least think they can).
Again, as the pace of these “notices of banishment” has picked up since late last month, it also suggests the Google has refined this piece of the algorithm to be clearer in detection of “unnatural” links. Let’s not rule out this is an unpublished Panda iteration refining how to distinguish “quality”.
How is Google Determining Unnatural?
I would suggest that it has less to do with where the links are coming from, unless of course, you are deliberately getting links from known bad neighborhoods. Then, you’re simply asking for trouble. As much as Google, Bing, or any other search engine on the planet believes they can dampen link signals in favorite of content quality signals and block-level analysis of sites, links still drive the web. As long as any website/web page needs a URI to be accessible, links will always have significance. Period. End of story.
You don’t have to stop getting directory links and other plentiful links. They exist precisely for those reasons, to give links to help people find your site. My belief is that Google is using a more technical analysis of websites’ link profiles: link velocity, linking root domains, and anchor text.
What is Link Velocity and What to Watch For
Link velocity is the rate at which links are acquired by a website. Link velocity is constantly fluctuating; there will be periods of higher link acquisition and periods where link acquisition will be lower. For example, your site publishes a controversial post/article, and the links come flooding in for a week. After which your link velocity will return to a more normalized rate. What Google might be keying in on is aberrant link velocity rates. That is, link velocity rates that are much higher than the normalized link velocity rate of the site.
For example, SITE X sees a sudden, tremendous growth of links over a one-week span, once Google has gotten around re-indexing all the sites. And, typically, SITE X has almost no link velocity. This could set off alarm bells that some type of manipulation has occurred. Whereas, sites like CNBC, a known and authoritative content publishing site, could see a spike in velocity several times a month, but its normalized rate would indicate there is always constant link acquisition, and therefore, is likely not link manipulation. Here’s an example of what non-normalized and more normalized link velocity looks like:
Linking Root Domains
Google might also be keying on this algorithmically to determine if your link profile is “unnatural”. What Google would be looking for is, relative to the number of inbound links a website has, a wide base of linking root domains and C-blocks to accompany those inbound links. Not only is that a measure of authoritativeness and diversity of the website (many different sites across the web linking to you), but it’s also great way to see if a site is piling up links from a few specific places on the web.
What this means, essentially, is that if your site has been stockpiling links from a few specific places on the web in a disproportionate way (i.e. you have several thousand links from X, Y, and Z and relatively low amount links from other sites across the web). Or, you have large clumps of links from specific C-blocks with very little diversity in your C-blocks, this is also indicative that there could be manipulative link building.
What’s the bottom-line here? Avoid big, chunky clusters of links from specific websites and specific C-blocks. If a site’s building links “naturally” it’s getting links from a variety of sources on a variety of C-blocks.
Anchor Text Concentration
Personally, I think this is the biggest key for Google’s algorithm to detect unnatural site link profiles. It’s something I saw last summer, and if you’ve been to conference in the last 6 months you’ve heard it, exact match anchor text on links for highly competitive keywords are trouble.
Let me clarify that by saying exact match isn’t to be avoided all together, but it has to be used judiciously and should remain interspersed with other semantic and temporal variations. Unfortunately, it is the past sins of link building that can be haunting you today.
I can only guess as to how Google sees it, but I would bet they attempt algorithmically to interpret how a “normal” web user would link to things. And, normal users don’t typically use keyword-precise anchor text when linking to things; they likely link with website/company name, a long string of words, short phrases like “here” or “this one”, or just simply the page itself with no anchor text.
The point here, is that when Google sees an over-abundance of highly-targeted, exact match anchor text with as much, or more, concentration than the company name or brand name, they have a pretty good idea that link manipulation has occurred. The bottom-line: if you are a new(er) SEO on the account, do a full-scale audit on the site.
Look at internal link anchor text (and whom you’re linking out to) as well as inbound anchor text. Group and measure the concentration of targeted anchor text for competitive terms, brand anchors, long-tail anchors, etc. If you’re seeing higher concentrations of targeted anchor text than your brand, then you might have a problem. Find ways to remove/limit the targeted anchor text and supplement with brand anchors.
Be Smart and Link as You Like
If you haven’t been officially told by Google that you’re unnatural, that’s a good thing. Count your blessings. But, that certainly doesn’t mean that they aren’t planning on it. If it were me and my site, I’d run a full scale link audit to make sure I knew exactly where I stood. Everyone can still link with impunity.
You can still get links any way you want, just be smart about it. Build a linking game plan: create lists of brand, exact match, temporal, and semantic anchors, the quantity each of those anchors will be used, and the frequency with which you get them. Even if you’re not “natural”, there’s no reason to pretend to be natural, so that you don’t end up on the “unnatural” list.
😀 Great post. When we link build for a client, we keep constant look out on anchor text thresholds based on % of root linking domains, % of total links, and % of mozRank passed by links with exact anchor text. I consider it a must in preventing OOPs
Regarding link velocity, I’ve always thought it was somewhat of an unfair penalty. For instance you create some sort of outstanding piece of content and it goes viral then you’ve just thousands of links (hopefully) within a few days. That would obviously make your link velocity jump through the roof. I don’t think you should be penalized or any alarm bells should go off for that.
I don’t necessarily think link velocity is a penalty on its own. Velocity is simply one measurement I believe Google is looking to determine manipulation. I really think that anchor text is key to how Google is determining manipulation.
If you have a piece of content go viral, it surely will cause a link spike. The problem is not sustaining link acquisition (i.e. after the spike, your site’s link acquisition falls off the map), and acquisition goes back to zero for an extended period of time.
Looking at it from Google’s point-of-view, a site that produces content should have some semblance of natural link acquisition constantly. That’s where I think normalization comes into play.
These days I no longer know which sites are good to obtain links from. Yesterday article directory links were cool but today it seems they are no longer too important.
Great post Tony.
You make some excellent points. The point about the different c class domains and different domain links in general is something that has been widely known/suspected for a long time.
Also the diversity of your anchor text, as you noted, is probably a fairly obvious trigger. With the anchor tex in particular, in my mind, it leads to the question, that if you are building links and you really diversify your link building efforts then naturally that is going to water down the amount of links you get for the specific keyword you are actually targeting.
Therefore, that does that mean that Google is going to be going back to paying even more attention to on page SEO again, so that it can safely determine the keywords it thinks you should rank for. I would be interested in your thoughts on that.
Yes, yes, yes, yes, yes. A thousand times yes! 😆
This is something I’ve strongly believed for a long time – at least a year, but more so recently. At the end of the day, it’s common sense: if every link to a site ever has the anchor text “blue widgets” – and it’s not in its domain, for example – then it’s going to set off a few alarm bells. But if there’s a big, broad mix of anchor text, it’s going to look a LOT more natural and therefore acceptable.
I agree with the link velocity theory as well, and even go as far as saying it probably ties in with anchor text. For example, I reckon that if you suddenly got a lot of links all with the same anchor text – e.g. “green widgets” – all in one go, Google might pick up on that, too.
The best strategy – as hard as it might be – would be to build links gradually and diversely, with a mix of anchor text and for keyword anchor text to grow at a rate that might not be deemed too fast (and therefore: unnatural).
I hear you – not had any warning on any of our clients but heck… i hear loads have.
Your link velocity is def one to watch but then so too is the amount of links you buy – like you need a 100 when 12 might do it guys?!?!
Ill throw a curve ball in – what says that these days a natural link building pattern isnt littered with spam – there isnt a site we propose for that hasnt been hit with some kind of spam directory submission or something as an attempt to call it link building…
what says you that spam now has a natural roll in an organic off page portfolio?
I say no – but fear that a yes is in there somewhere!
nice read tho – lays it out – I would add – dont reply to the messages if you have got one. bad idea to confirm that you have any control over ho links to you site – heck you just ade the site to serve users right? right!
Good post thanks for sharing Ant!
Do you think the penalties for link velocity and anchor text are more industry specific? I try and stay away from this personally, but I see a TON of competitors getting away with it and they are grabbing hundreds of thousands of new links every month or so with almost all the same anchor text and seem to be successful.
Additionally, do you think if a site has a spammy link profile, but isn’t a spammy site or garbage site for its query that Google may ignore this signal because user engagement via the query is strong?
Comments are closed.