Detecting “Undetectable” Link Networks
I have been growing more and more concerned lately as clients have been coming to us with propositions from link builders who claim to have an “undetectable” link network. Over time, link network operators have become more and more sophisticated at creating more prolific and difficult-to-detect networks. The process, while straightforward, always seems to be the same.
- Acquire Dropped or Pre-Release Domains with Existing PageRank and Strong Link Profiles
- Host Domains on Different Class-C IP addresses
- Put up unique content with links
Unfortunately, this creates a pattern that is difficult to detect for even moderately trained SEOs who lack the technical skills to analyze the myriad factors that Google has at its disposal. Subsequently, a human looking at several sites on the network would see few if any patterns – the sites look and feel very different, the registration data looks different, they all have different backlink profiles, etc.
This creates a dangerous situation for clients who are now presented with a cheaper alternative that at face value seems to be risk averse. This, however, is rarely the case.
There are tons of factors that Google can look at to determine patterns among backlink profiles that would help uncover a link network…
- Registrant Company: Sure you have domain privacy, but all of the domains have the Registrar
- IP Owner: Sure you have different Class-C IPs for every site, but they are all still HostGator owned
- Platform: Sure you have different designs, but everything is still WordPress
The truth is that mixing either of these with intersecting backlink patterns is enough to nail the overwhelming majority of private, for-hire link networks on the web.
Example Link Network
A client of ours presented me with a solicitation he had received just a few days ago regarding a particular private, for-hire link network. He requested to see a couple of the example sites, which is standard protocol, and he provided me those links to determine whether or not the network actually stood up to the claim that that it was “undetectable by Google”. Surely if Virante could detect it, so could Google.
So, I dropped the 7 example network sites provided into a tool I had written that looks for backlink patterns. It first found all the sites those network examples linked to, it then found all of those sites backlinks, then checked all of them to see where they linked, so on and so forth. With only 2 iterations deep plus adding in a WordPress footprint detection, we were able to conclusively identify 118 network sites and 31 users of that private network. The spidering and calculation took our servers about 5 minutes to complete, and only employed 1 of the additional metrics we can use to identify networks.
So Why do Private, For Hire Link Networks Tend to Suck?
The primary issue is economics. Let’s say you truly wanted to create an incredibly difficult to detect link network, here are all the steps you would need to take…
- Acquire domains from numerous sources: SEDO, Godaddy Auctions, etc. (assume each domain on average costs you $70)
- Transfer to random registrar with privacy (assume $12/year)
- Acquire hosting on unique class-c IP on a different with random hosting companies (assume $5/month = $60/year)
- Acquire uniquely written content for site, with random number of pages (assume average is 4 page site, at $5/article = $20)
- Acquire unique design for site not using same platform (WP, Drupal, etc.) over and over again (assume $5/ site)
Although this strategy still has a huge footprint of having domains sourced through brokerage websites, the above model would reasonably cost around $170 to set up, and then would continue to cost $70 a year to maintain thereafter. At this price point, nearly all link networks need to sell at least 2 links off their site to make a profit. When this occurs, it becomes susceptible to the simple backlink analysis strategy discussed above. To avoid this, they would need to provide only 1 link per site which, at that point, makes it more cost effective to simply source links via targeted link building rather than to build and maintain your own network.
38 Comments
Trackbacks/Pingbacks
- Canadian SEO | Local Canadian SEO | Hamilton SEO|CanuckSEO.com - [...] run by a Russ Jones who is a Virante.com CTO. Never met the man, never heard of his firm…but…
- Różnicowanie parametrów linków w procesie Link Buidling • ArtykuÅ‚y • SprawnyMarketing.pl - [...] …to kilka pierwszych i podstawowych elementów o których wiÄ™kszość twórców prywatnych sieci blogów, precli lub sieci katalogów zdaje siÄ™…
- L'actualité SEO vu par SEOmoz - Moz Top 10 09/11 - [...] Détecter les réseaux de liens peut s’avérer très facile grâce à des éléments basiques tels que propriétaire de l’IP,…
Nice post, thanks for divulging.
One problem that I have is that some of my clients don’t really understand, and to be honest dont really care care about link quality. as long as they rank page 1 for their desired terms.
How do you advise your clients to stay with yourself and not to change their linkbuilding campaigns to these link networks?
To be honest, we lose a lot of those arguments. But the clients who do stick around see great, permanent gains.
We get a lot more of these types of offers than our clients do and they always fall apart when we get a sample report of what they offer. The domains are often poorly generated, obviously fake companies and/or expired domains that have been re-purposed.
While we have yet to entertain one of these offers, I imagine having a tool to help verify and double-check their quality would still be of some use.
The biggest problem we run into is that our client’s competitors often do utilize strategies like this and it works for them (short and long-term). As Richard elicited in his comment – How do you convince your clients that this is not a good strategy & to follow yours?
Russ – Is the tool you referenced to help look-up this network available for sharing? Purchase? Etc?
The tool is not available at the moment because it is expensive to run. If we start with 5 example network pages, and they each link to 3 sites with 1 crossover on each, we have 10 domains backlinks to look up. We then query the top 1000 links from SEOMoz Site Intelligence API to find crossover, then we repeat the whole process with the new batch. It ends up calling tons and tons of data which is both $$ and bandwidth prohibited to scale outwardly. Maybe sometime in the future, though.
Thanks for the detailed response Russ – much appreciated. Great set of tools btw… we need to use them more frequently! 🙂
These proposals get more and more concerning, especially with the recent tendency to even make it sound like Matt Cutts and Google are endorsing their methods!
One lesson to be learned…if you lose a client to one of these companies, keep their file where you can find it. Chances are they will be back before too much time passes – unfortunately, there may be a lot of repair work to be done when they return…
Interesting article. I own several of the link networks mentioned in this article and service over 100 clients with them and other methods. Regarding your detection tool–every factor you mentioned can be minimized and avoided by careful planning. Additionally, even those who are too lazy to set up their network well enough to avoid detection using a tool like you have developed, for whatever reason Giggle hasn’t made it a priority to stop them. An easy way to verify this is see if the lazy networks you found are still indexed, and where the sites being linked to are ranking at. Almost always they’ll still be indexed and the rankings will be quite high.
All of this makes for good theoretical talk, but for those of us in the trenches battling for #1 rankings for massive huge money terms, you do what works, and that’s that. If Giggle gets around to detecting these a few years from now, I’ll be too busy sipping pina coladas on my private island made of stacks of cash to care.
Your 3 criteria listed essentially label any wordpress site on Hostgator as a member of a link network. You and I both know that’s absolutely not the case and Google is smart enough to figure that out as well.
Now as you mentioned, linking patterns can quickly give away a link network but I’m not convinced Google puts all that much effort into detecting or devaluing those. I can show you plenty of examples of sites that rank only due to a network of very easily identified sites.
Now I would never recommend relying solely on any one type of link but I thinkany SEOs are a bit too cautious and give Google a bit too much credit.
The ROI simply isn’t there for most webmasters to build and maintain their own networks. However, leveraging other people’s networks at a minimal cost is a no-brainer if it works for them. I could show you some blatant networks 1,000s of sites deep that are ranking for highly comp pharm terms…When it comes to cutting edge stuff, these are the SEOs to watch imo.
I’m surprised you would refer to these networks as black hat. There isn’t anything black hat about them, grey hat maybe. I suppose it depends on how they are built and used.
I just got done building my own network and I’ve taken every precaution to avoid detection. B-class IPs spread across the world, multiple CMSes, private whois, custom name servers, 100% unique content, no interlinking anywhere…The only thing a private network needs to do to avoid external link patterns is drop links on a percentage of your sites in the network instead of all of them.
Ah ok that makes more sense, I thought you meant those 3 criteria were the only thing needed to find a network.
And yes, I would agree that your money is usually better spent elsewhere although there are times and places for just about every link resource 😉
I’ve seen the link graphs that “legit” SEO companies build (including SEO companies endorsed by SEOMoz), and trust me…it’s not always pretty. So, that mumbo jumbo about “white hat” techniques being better or less risky is not going to win this argument. When your client is a big name brand, and you’re ranking them for their niche keyword…I’m sure slow and steady will win the race. However, like FYI King said, when you’re dealing with very competitive spaces, it’s any port in the storm.
This claim that what a “white hat SEO” company does is so much better than a link network solution, is crap. At least the client understands the risk going in to a link network, and we’re not dressing a wolf up in sheep’s clothing. The risk of Google detecting a “link network” is just as likely as the New York Times discovering a shody link graph built by a “white hat” SEO company. Having seen the work that Virante does, this holds less true for you. However, there are big name SEO companies out there who are doing anything they possibly can to hide the links they have built for their clients.
So your whole argument is that your money is better spent ‘elsewhere’? Maybe ‘elsewhere’ being your company?
Oh and and Google will surely….maybe….possibly in the future’catch’ you writing writing unique articles and getting them posted on a blog that was so easily detected because it was used wordpress, was hosted through hostgator and registered through godaddy (hmmm sounds like billions of other sites).
lol.
First I would like to say thank you for bringing attention to the subject of spamming a website–link network or otherwise. That will never help in the long run to rank for anything.
On the other hand–as you have stated in a few comments above–using a privatized network that is careful with what they are doing, and creating their own niches in a network that certain links can go into, is not hurting themselves. I highly doubt that Google will take the time and effort (according to your comment about not having the tool anytime soon) to find the handful of this or that network. If they do write something then it will either cover everything or only the few that are doing it wrong. I never thought I would hear about the day Google puts down relevant and 100% unique content to any site or niche.
So far link networks (and by that I mean the ones doing it right) are the closest thing to getting along with Google I’ve heard of. They create sites for Google to serve their ads on. Not only that but the content is unique and targeted to what a consumer would want to hear about in any given niche. Instead of spamming the internet, good link networks are finding a way to work with Google.
Private link networks are risky. I always make sure I don’t link out to any of these sites. Overall… while it might help you in the shortrun, it is not a long term strategy. I don’t rely on these networks very much.
A lot of SEOs will also just acquire the links for their clients and not actually test if they work. So many links now get discounted by G’s algo.
What do you think about link networks that require the users to submit their own websites instead of paying?
That eliminates the IP address problem and gives you an insane amount of linking domains. I’ve been using this type of link network for a few months and it’s a really solid solution to this problem. As everyone else has said, it’s obviously not smart to rely on one type of backlink but it’s one of the most effective in terms of effort:results.
Completely agree with you. No such thing as an “undetectable” network.
The reason I see for Google not slamming down on networks in general – it’s an incredibly “gray” area. Sure they can be spotted by the general footprints:
– WordPress site
– Same registrar
– Same “mash” of content
– Same IP owner
But the problem is, there are millions of sites just like that, that aren’t in networks. The link network sites hide by being “part of the crowd”. The only differentiator is that “normal” sites tend to focus around central themes, while as you mentioned, for the link networks to be viable, they have to accept links from just about anywhere.
I think AI is getting better and better, but it’s still not at the stage where an algorithm can visit a site and know FOR SURE that the link to “Mesothelioma lawyer” isn’t actually a normal site owner linking to something they are dealing with, when they are a stay at home mother and usually blog about the latest “stainless steel cookware” they just bought and the “carpet cleaning company” they used the previous week.
And that is why Google doesn’t do anything about them. For now.
We have been getting a ton of emails from sites like this. How the heck are they finding us? Also, it becomes a little scary because we have legitimate partners that we link to who we have caught participating in things like this. How can we protect ourselves?
have you done any research like this on small niches? I would think that this type of footprint/interlinking is present in pretty much any small to medium niche where there are only 100 to 1000 legitimate sites in the whole niche. Take a particular video game’s niche, for example — First Person Shooters. All the fan club forum sites will be pointing at all of the popular first person shooters, and all the fan club forum sites will be pointing at each other.
it seems that even if you can decide that some group of sites is linked together, you can’t reasonably conclude that it is a private network.
Sorry if i’ve missed it but not seen much mention of the benefit of building ‘genuine’ links on traffic and conversions. Some of my clients get nearly 20% of the their traffic from the links we build and that traffic converts. A link network is unlikely to send any real traffci that converts to sales.
The thing with this approach is that it can work extremely well and it is not penalized. What you’re lumping together is a high quality use of networks with spammy use of networks. If you are posting auto-generated crap or articles written for $1 per article, sure, it’s garbage and that’s what Panda addresses (and future updates, I’m sure). But if you are putting out high quality content, well written, grammatically correct., etc., with a well-focused and thought out topic, then I think it is highly relevant and, in fact, Google rewards it.
I agree with the title : “Undetectable†Link Networks”, which means “undetectable as long as no one investigates further/deeper”.
Loads of linkwheels and other mininet schemes are available. The story is always the same to spot on private networks : too few different metrics. One thing you didn’t mention is the date/age factor (ie. new websites linking to older ones).
Regards,
you mentioned de-indexing and reinclusion requests as a reason to to participate in a link network. Let me ask you this:
Lets say that G does track down and identify a network, and slams those nasty link spammers who participate in it, and makes those sites “poisoned donuts”. Then some competitor adds your site to the network – what happens to you? Re-inclusuion request anyone? Tell them you didn’t put your site on there? someone’s out to get you? Sure, they’ve never heard that before. Oh, and thats assuming you can figure out WHY your site suddenly plummeted out of the SERP’s for no reason you can think of. G is notorious for penalising first, giving reasons later (if at all).
Non participation in link systems does NOT provide invulnerability from a G pogrom should it happen. All that would achieve is a temporary sting of link spammer sites followed by a repositioning of that network as a link landmine and wholesale slaughter of opposition sites, and hiring out as an SEO Assassin network instead of a rank building one.
I really hope G doesn’t take your advice, for all of our sakes!
Do you think G looks at G Analytics, G webmaster tools and Adsense pub id to group websites as belonging a single owner?
I could argue about the costs you mention for maintaining a network as being way too high and I could also argue about a good number of the other points you make but you have already clearly stated yourself why this sort of network will continue to help sites rank.
Your network detection tool is not available at the moment “Because it is expensive to run†and “It ends up calling tons and tons of data which is both $$ and bandwidth prohibited to scale outwardly.â€.
I’m quite sure it is.
Google clearly has an enormous amount of resources at their disposal but it also attempts to gather, process, interpret and update an amount of data that is almost unimaginably large. The reason why their algo often looks so incredibly easily gamed and naïve is a matter of scalability. Of course they could build a tool like yours and find link patterns everywhere but then to roll that out across the network…you say you can’t do it and neither can they and that’s why a carefully crafted networks will continue to rank sites.
This is moot. I’m assuming the sites in question are, although unique, still are low quality and not many of any people are reading / using them? The give away metric is the usage stats for those “only for SEO link building poorly written” sites. When the page gets no reads, and 100% bounce rates, the links on it become either null or even bad link juice. Too many of those pages and the entire site tanks.
And if people are reading these sites and using them / sharing them on social networks, then these are not sites strictly for link building. They are sites.
One point repeated a number of times on this page is that the farms start with buying domains with an existing profile via auction. Why is this not a detection metric – surely Google can monitor the domains available for auction and factor that in?
Registrant Company: have multiple account in different registrar/whs …
that’s a new market for seo supplier …
they will purpose different class C ip, private whois, and diferent registrar … for only few bucks more !
IP Owner: same method as previous …
Platform: french web hosting services allow user to one-clic install joomla, wordpress, dotclear framework !
Sebastien
with $170 i think i can do much more !
I am thinking of creating my own network. I have managed to rank very high on a very competitive keywords. As far as you keep link building and don’t relay on your network then your site will rank high. I know that guys from google are trying to make a lot of cash on expensive google add words and I don’t blame them for that but if you want to rank high on competitive keywords in a short term then there is no other way then hiring a private network.
As mentioned by one of the guys on here; imagine that your competitor would put you on some assassin private PR network. What would you do? I could get rid of my competitors in a week. This is one of the main reasons Google must consider before they kill your ranking. I have seen some major companies like hotwire, hotels.com paying for high pr8 and pr9 links and google doesn’t do anything about it. The funny thing is that you cannot even track these links with seo spyglass 🙂
You’re all only looking at one aspect of SEO. There is so much more, and more important things to consider. Only concentrate a small portion of concern for link building. There are other more important aspects to focus attention on. Only worry about high PR links you want them to be legitimate. Start building link wheels from high PR sites and use those to interlink with your other site wheels. Focus on legit social networks. You’re right. Make more presentations, and videos! Legit blog sites are a great tool for SEO, as are music sites, and if you slip in a few .edu links you would be amazed at what it will do for you!
also…
Don’t go through this…
“Factor in the cost of a Reconsideration Request: How much will it cost to remove all the links, go without rankings at all for 2+ months, and file multiple reconsideration requests”
I can tell you an easier way to get around this problem. I was sandboxed by Google because a client didn’t get me the Google Places code that they mailed to them… I went from the second position on page 1 to non-existence. When the problem was corrected I was booted to page 16 from page one. Within a week I was back on the first page. The fastest way back is to completely re-build the website at least the html. Change the title, and all the meta data. If you really want it back fast put up a wordpress page and load it down with your keywords and tags! As soon as it jumps to page one go into your hosting account and remove the link to WordPress and install your real website. I know it sounds crazy but it works and it sure beats groveling at Googles feet for 2 months! Lol
By the way I applied for your job opening!
So lets say that I wanted to build my own network, I’m not an agency. I’ve always believed if you know SEO you do it for your own projects and let everyone else deal with ‘clients.’
If I wanted to avoid detection I would do all the obvious;
– unique name servers
– different IP’s
– different WP themes
Then I would add the following;
– I would find a creative way to mesh the old content with my target keywords.
– every 50 sites or so go with different hosting companies
– 75% of outbound links go to high authority sites, .edu sites, government agencies, national geographic etc…
– each network site only links out to 1 site I am monetizing
Doing this is costly and time consuming, but at the end of the day I could defend the content of all the sites in the network, and all the links.
What are the detectable holes in my system here I am vulnerable to? I’d like the stacks of money and pina colada’s too, but I’m just starting out and have only 1 stack of money and just a drink umbrella.
Thanks
The trick is surely nice. I will try it on all of my clients sites without any doubt. Link networks actually sucks and another part of that is Google gets those site penalized for the same.
This was really interesting and I’m interested in reading some more. Do you have a similar piece on link exchanges like Voltrank? I’d be really interested in seeing how susceptible to being recognised those services are.
That’s ignoring the fact that using Voltrank often forces you to link to none relevant websites which could be using some horrible methods and effect you badly because google looks down on them for their spammy ways. Stock junk sites and the like abound. So many of them are all trying to run a quick scam.
Very nice and informative article. I stiil think however that well planned network can work good. Obviously it is expensive but gives You independence and at some level can lead to more stable serps.
Hi,
I have to totally disagree with you. Your “footprints” are registar, IP owner and platform – and they’re kind of irrelevant.
Let’s deal with registrar first.
http://www.webhosting.info/registrars/top-registrars/global/
Apparently Godaddy has 31% market share for registrars. I really fail to see how having all your domains with Godaddy would be a footprint Google could find. That’s just silly.
Second is IP Owner/Web Host
Here’s another sweet one for you…
http://www.stackify.com/managed-hosting-providers-market-share-visualized/
Rackspace has a solid 5% of managed hosting marketshare. Given that most “hosts” nowadays (the ones I’ve seen) are really offering just little chunks from the large hosts (rackspace VPS broken up) – even if you have 5 IPS with some random mini host the only information on that IP with Arin or anyone else is that it is a Rackspace IP. Again that’s 5% of the world. Think you might be safe blending in with a couple dozen million other sites. Hardly a cutting edge footprint – or any kind of footprint for that matter.
WordPress as a footprint. You’ve got to be kidding. That’s 20% of the entire WWW that is running WP and that’s supposed to be a footprint?
None of what you have listed is really a viable footprint in any way shape and form. What *might* be a viable footprint from what you’ve identified is…
WordPress plugins – If every site has the same 8 plugins it’s pretty obvious. There are also plugins that hide what you site is running. Just takes a few minutes.
Themes – don’t use the same stuff please. Though you could hit the top 10 themes (with 2 mil downloads on the top 1) and be pretty darn safe.
Footprints are something that link YOUR sites together. A footprint like using godaddy links 30% of the internet together. That hardly qualifies as a footprint.
That’s my 2 cents worth 🙂 Aside from this post I’ve been pretty much amazed at your site!
Though in your defence, a *PAID* network would hardly care to watch for footprints. They are just after money. However having your OWN network you know to take care.
That was a quick reply. I totally agree with you on that. The biggest footprint will be 50 sites and each one links to all 5 of my money sites. That would be a huge foot print. Limiting outbound links is a good idea as you mentioned.
On a side note, do you work for Virante?
The author of this post is not trying to give you a complete list list of factors that could potentially help Google identify your network. He says there is a ton of factors and i think he speaks literally. So you own a blog network… did all the ip, registrar, platform, skin, content stuff, but forgot to not google every domain from the same computer. Whoops!
But wait.. Greg says WordPress is about 20% of the world and Rackspace is 5%.. No footprints, yeah? But a network of 80 blogs, all on wordpress, has a chance of 0.2^80=1.2*10^-54%. I guess you should feel special 🙂
I want to congratulate the author of this post. I am sensing that most people think the author’s argument is a theoretical one. I think it is the most practical i have read in a while. More than 2 shared links make a network identifiable to anybody having access to the link graph. If your setup requires more than 2 links to be economically sound, then it can be identified. It may be that it takes some time to identify it, or that it is not economical for Google to identify it (i think they are), but the algorithms to make it happen are already there and they are quick.
If i were Google, i would identify blog networks by first digging out unnatural link patterns (where 100 blogs link to the same 20 sites for example), then checking for blog network evidence such as platform use, ips, etc. Google can certainly do that in its own time, when it can return a list of 1000 sites in a split of a second for the most obscure keyword phrase.
Yes, yes, I know blog networks work. They have been for over a decade. If Google decided they stopped working in one day, the internet would get “lost” and Bing would be happy.
but the algorithms to make it happen are already there and they are quick.