Matt Cutts has more in store for us webmasters….

Toxic Link Cleanup Phase

Over the past year or so, webmasters began getting bombarded with “unnatural link” warnings in Google Webmaster Tools. Understandably, this caused somewhat of a panic, and webmasters began making frantic efforts to clean toxic and suspicious links from their sites’ link profiles.

These “unnatural links” had become a much bigger problem than most webmasters even realized — after running more than 100k Link Detox reports (which are designed to help webmasters identify and disavow unnatural links), we were surprised by how many really bad links were out there. Google truly did need to implement some way to force us, the webmasters, to clean up the link graph mess.

Many webmasters have reported successfully revoked manual penalties after cleaning up their link profiles and submitting their sites for reconsideration (typically a month-long process). Some have even experienced penalties being lifted after simply disavowing or removing enough links; doing so lifted automatic filters without the need for a manual reconsideration process — which is great.

But what’s next?

Enter The SEO Risk Management Phase

Unfortunately, recovering from a Penguin penalty once doesn’t immunize your website against future penalties.

In fact, as I pointed out last year, many of the penalties imposed by Penguin were a long time coming — and this suggests that there’s a lot more to come. Thus, webmasters must be conscientious, actively monitoring their sites’ ongoing link growth for anything that might be construed as “spammy” or “unnatural.” While a suspicious inbound link may not trigger a warning now, that doesn’t mean it won’t in the future — all Google has to do is flip a switch.

Risk management for your links is both an ongoing process and a mindset. While there are programs out there that can help you identify “risky” links, running these programs and disavowing the suspicious links is by no means a “one and done” kind of deal. We’re not talking about a singular event where you clean up your old sins and are safe for the future — risk management for SEO and links entails assessing the risk of potential new links, as well. After all, if you have acquired “bad” links in the past through active link building efforts, you don’t want to risk additional harm to your site by continuing to build similarly “bad” links.

A Link Is A Link — Not

The big shift in mindset here is that the old saying, “a link is a link,” is simply no longer true. Search engines have increasingly put measures in place to evaluate not just the quantity, but the quality of inbound links, meaning that an inbound link profile consisting mostly or entirely of these “low-quality” links triggers a red flag. What’s more, a link that helps one website could hurt another one. Some examples of this include:

• The website being linked to already has inbound links from the same network or owner.
• The sites being linked are thematically unrelated — in other words, one website is linking to another website that is not thematically relevant. For example, a “Buy Viagra” link to a tax adviser website is problematic; whereas, it would be considered thematically appropriate if linking to a site where one could actually purchase Viagra. The same is often true for other verticals, but the “porn-pills-casino” (PPC) links are specifically bad for most sites.
• The link anchor text is too keyword-heavy. While it is acceptable to occasionally use keyword-rich anchor text to link to your site, it begins to look unnatural if the overwhelming majority of your backlinks are like this. A healthy backlink profile contains a wide variety of  links, including homepage links, subpage links, links with missing anchor text, links with “Click Here!” anchor text, links with keyword-rich anchor text, contextual links, links with the brand name as anchor text, etc. If your backlink profile leans too heavily toward keyword-rich anchor text links, this may trigger a red flag.

From these brief examples, you see that there’s one important thing that’s changed with the Google Penguin updates: you have to look at your link profile as a whole and review each new link in the context of your existing links.

Do You Overlook All Your Link Campaigns?

Julie Joyce wrote about the issues that occur when clients hire multiple link building companies; and I can confirm that dividing link building responsibilities in the name of diversification has the unwanted side effect of “diversifying” the responsibility for bad links.

For example, if client X hires three link builders — we’ll call them Tony, Fred and Susan — to build X links per month, s/he’s already in deep trouble if s/he doesn’t review their plans before they actually build those links. Chances are very high that each will secure link spots on websites that are somehow related (especially if we are talking paid links), because those are often sold from the same networks and/or same owners to many agencies.

Thus, the process in link building often goes something link this:

1. Client X tells Tony, Fred and Susan: “Get me 20 great links!”
2. Tony, Fred and Susan reach out to their network for quick/paid links for links that are thematically related to Client X’s website. Of course, those are often unnatural links because they are paid or swapped, but let’s assume for a moment that they would be so well camouflaged that Google could not detect it as such (rare, but possible).
3. These webmasters separately respond to Tony, Fred and Susan to present them with potential linking opportunities.
4. Tony, Fred and Susan each pursue the strongest links available and manage to get Client X linked there.
5. The result is something like this this Venn diagram:
   

   Risk for your SEO happens when link builders’ plans are not reviewed and controlled properly.

   As you can clearly see, these three link builders would inadvertently work to create a very unnatural backlink profile, due to a lack of communication and oversight. And, we haven’t even taken into account the thousands of links that Client X could already have! Can you imagine what further conflicts might exist?

   Performing What-If risk checks for new links

   The solution to the problem of overlapping link-building efforts would be to carefully review all link opportunities in advance, prior to actually building the links.

   Every webmaster should therefore do the following:

   1. Review all existing backlinks for toxic and suspicious links, and note any that you plan to disavow or have removed.
   2. Examine your link builders’ list(s) of potential link building opportunities.
   3. Check to see if the links your link builders have identified are coming from the same IP network, same domain owners, or same Class-C blocks as any of your existing back links. (This requires gathering domain information for every potential link from their respective data sources.)
   4. Create an overall SEO risk assessment for the potential links your link builders have identified.
   5. Decide if that link is worth getting based on its risk for your site.

   While gathering all those metrics for potential links and checking them against your existing links can be tedious, it’s mandatory in a post-Penguin world. Link Detox eases that process with a feature called “What If” Checks, and other products will likely offer similar functionality in the near future.

   When doing such a simulation, make sure you ignore those links that you potentially disavowed. Keeping track of previously disavowed links is another story and a crucial detail in an ongoing link risk management. You would have to remove all disavowed links from your profile, just as you assume Google does to not interfere with your audit results. We also call “Round-Trip Disavow” in Link Detox, and I’m looking forward to seeing other products adopt that crucial methodology.

   Better Safe Than Sorry — Conclusion & Proposal For SEOs

   The above example is just one of many issues that can arise from a lack of proper SEO risk management. Every webmaster must be engaged in the link building process, and they must understand that every potential new link poses a risk — be it today or tomorrow.

   I recommend the following solution:

   1. Review your backlink profile regularly for “risky” links. You can automate that with software that generates recurring reports; creating a manual process will work, as well.
   2. Perform ongoing link cleanup. If you find fishy links in your backlink profile, get rid of them ASAP — don’t wait until Google “asks you to” by slapping you with a penalty.
   3. Keep contact records of all link sources. Whenever you do receive a penalty or warning, you should be able to react just as quickly as Interflora.
   4. Keep your contacts “warm.”  In other words, ensure that you can easily reach out to any link sellers/traders you work with in the event that one of your backlinks needs to be removed urgently.

   Some of the above might sound like recommendations for link buying or unnatural link building. But in fact, these are just typical examples of what is (still) going on in the SEO industry — especially putting link building tasks into the hands of many companies.

   Having reliable risk management processes for SEO in place will go a long way in future proofing your site(s) against future Penguin updates. Be proactive and start putting your process together right away — don’t wait until it’s too late!

   (Disclosure: Link Detox is a product of CEMPER.COM)

Now, just this past Sunday, a week before Mother’s day in UK, they are already back for most of their rankings. Many wonder how that recovery was possible so quick and we’ll look at what was changed and what was cleaned up.

More Than Advertorials

As we could see in the Interflora Deep Dive Analysis the 100 or so Advertorials where not the only bad links in the mix, albeit Google specifically “reminded” about paid links in Advertorials being bad as well.

Interflora actually had more than 70% of their links as toxic or suspicious according to Link Detox. So those were actually 1000′s of links that would probably not pass a manual review by the Google spam team.

If you look at some of the examples of Tiered backlink structures (aka Linkwheels) or the huge number of low Power*Trust links compared to the competition you will wonder just like so many others why Interflora actually wasn’t hit before in the Google Penguin updates, as such backlink profiles were clearly those of the losers.

They had a lot of free blogs with simply one page on it, carrying a money keyword like i.e. “roses” to their site or to another simple free blog. That was a common link pyramid tactic implemented with link automation tools.

You will probably also wonder how they got back so fast, just in time for Mother’s Day.

Link Removal & Disavows

Interflora losing rankings even for the brand name indicated a manual penalty by Google.

The manual spam penalty and news around the globe already indicated high level attention at Google, and while Matt Cutts didn’t specifically mention Interflora, the correlation of events seems very plausible.

We also heard about a lot of bloggers being contacted to remove their links to Interflora from various pages, which many of them found a good thing to do. Some said that the guestposts or “paid” posts were low quality content anyway.

This outreach was so big that the whole industry noticed, including Google. Something they probably wanted to happen to make Interflora to an example.

We also saw that advertorial links were quickly removed from regional newspapers already in a first analysis following the penalty.

Interflora changed their link profile to contain 10822 links (from 10944, so only minus 122 links) but the number of pages they got the links from increased from 7276 to 7442. This means that actually a couple pages that had two or more links are gone now.

One example is the following very obvious paid blog post, which could still be found in the Google Cache and had 3 links to money pages with money keywords and the Interflora brand in it (a common tactic we call “Compound keyword phrases” that is used to make links pass money keywords  camouflaged in variations with the brand in it as well).

Now while a lot of bad link building examples were shown in the Deep Dive, this is another poor one, and a quick look at the site reveals this pretty spammy sidebar:

and in this (broken) contact form:

So as you can see immediately, this was not a great link either, so good it’s gone.

To answer a typical question that comes up at this point when we look at results of Link Detox – the number of links were not inflated by sitewides, which means no more than 5 links per domain were counted.

This post explains more detail about the reason; but in general – 1000s of subpage links are no valued as much as 1000 domains linking to any site, as was just recently confirmed by Matt Cutts. The inflation of link counts by sitewides is a big flaw in many “backlink checking tools” and often leads to misinterpretations by webmasters.

For more inspiration on which good links Interflora could build now, the post about 101 Links for Interflora gives more ideas for how they could proceed now.

But the problem is, even today the link profile of Interflora shows a lot of suspicious and toxic links in Link Detox. The number of healthy links went up from 2164 to 2207, but still exactly 29.7%.

It’s worth nothing that also not all advertorial links were completely removed, but the remaining three will probably be as you read this.

Now while we see that a mere 122 links were actually removed based on this quick analysis, the questions we cannot answer are:

• How many links could not be removed and were thus “disabled” using the Google disavow tool?
• How many links were disavowed because contact information is unavailable or lost (like maybe for free host spam)?

Why these question? The number of “disabled” links is probably a lot higher than 122.

How The Disavow Process Works

Google recommends that a penalized website should try to remove every bad link possible, and disavow the others that could not be removed. Disavowing links works by uploading a simple text file with links that a webmaster wants Google to ignore. They take it as “a strong recommendation” in the next crawls to ignore those links.

This is the interesting part – even if Interflora would have disavowed thousands of shady links in the last week, the recrawl for all those pages could have taken weeks or even months. This is because Google doesn’t give any indication when Disavowed Links are discounted. Also many of those bad links came from pretty weak pages, which are not recrawled very often naturally anyway.

At this point, we come to expect that either:

a)  The recrawl process for Interflora links was really sped up by Google (manually)

b)  The mass of spammy links were already disavowed in the past as part of a Link Risk Management undertaken, which would make a lot of sense

c)   The manual penalty was not only applied but also revoked, and included “forgiveness” for a lot of these other bad links. Matt Cutts himself maybe pushed a button to get the links recrawled faster, so disavowing would work faster.

We can however assume now that in addition to these 122 removed links a lot more were disavowed.

External Link Profile vs. Internal Link Profile

This brings us to the point often overlooked, even in the previous analysis pointing out 70% toxic links for Interflora. The backlink profile that any software can show for a website can only be accurate if we assume that no links were disavowed.

As soon as links were disavowed (and recrawled) Google would probably not count them anymore, and neither should we.

Therefore only Interflora and Google know the answer to above options, and ever link analysis is being diluted by links that Google maybe already has discounted because of disavows many weeks ago.

Using the “Round Trip Disavow” in Link Detox webmasters can allow Link Detox to discount links that were disavowed as well, just like Google does. Thereby one can keep track of its cleanup and disavow process and get a more accurate picture of his backlink profile.

Discounting Disavow Links Too

The file listing all disavowed links can also be uploaded, and to demonstrate that, we’ll do that really quickly by picking 6981 links from really bad sites, virus sites, sites with Page Rank penalties and several more reasons and uploading that into a Link Detox analysis.

By this, we get a picture of only roughly 40% problematic links remaining, which probably comes a lot closer to what Google currently sees. Still a lot, but every website has some problematic links. It’s the ratios between good and bad that matter.

Keep in mind only Interflora and Google know which links were disavowed actually, this should only serve as an example here.

Summary

Given that the outreach to bloggers was done so fast and wide-spread, we can assume that Interflora was well prepared and probably already warned about Unnatural Links a lot earlier. This makes a lot of sense for Interflora and any company depending on Google’s organic traffic. Link risk management is something that came with the Google Penguin updates and the first time that poor links can cause harsh traffic drops and penalties.

We might also assume that:

• According to this case study on Unnatural Link Warnings and their effect Interflora probably just did not trip the threshold for traffic loss earlier
• But the warnings probably already led to preparation of outreach lists and previous link removals by Interflora
• Previous link disavows made a lot of sense, given the links we saw they still have on free blog hosts
• The ashamed Advertorials were the tip of an iceberg that were pretty much overdone – causing manual intervention
• Interflora and/or their SEO agency reached out to the Google spam team or Matt Cutts right after the penalty to declare more sins and disavow more bad links, also show action by their immediate reach out
• Google got the reminder on paid links in Advertorials spread across the world really fast
• After global press coverage of both the Interflora penalty and the most prominent paid link example penalty since JC Penney, was another successful “PR stunt” as some call it by Google. The timing here is also very similar to the famous JC Penney penalty. That one also hit at a time that meant very bad PR within the SEO industry, but “minimal” impact in terms of their revenue, right after Christmas.

Assuming everything went fine for Google, all poor links were disavowed or removed by Interflora there was no reason to hurt the user experience of Google clients – their users – anymore. This is very good for Interflora’s business of course as well.

The unfairness that many see is that Interflora gets back so quickly, when smaller websites struggle for months in their cleanup and never seems to get released from the penalty. After all, someone Googling Interflora would expect just that – Interflora, and not a lot of SEOs and Mainstream media writing about their penalty. Unless it’s an SEO Googling…

That’s the beauty and power of being a big and important brand. Google just cannot have it in its search for long, otherwise “Google would be broken”, as my mother would call it.

For a smaller brand that’s bad news, as such fast recovery then would not be possible. This means everyone should start with link risk management as described above, today.

For quite a while, it was possible to rank in the TOP3 with literally no backlinks, compared to non EMDs, often after only a few days in literally every niche you can dream of.

Exact Match Domains (AKA Keyword Domains) are, in general, domain names that exactly match the keyword a website wants to compete for. For example, if a website wants to rank for the term [minor weather report], an exact match domain would be [MinorWeatherReport.com].

For years, it was a free lunch for those in-the-know, and now this loophole is closed, just like other loopholes have been closed years before. But instead of complaining about a free lunch being taken away, SEOs should be thankful for having had it. This cheap way to get traffic, thanks to Google being pretty slow to close the loophole, is now gone.

What The EMD Bonus Included

Two of the first SEO techniques that are taught are on-page keyword optimization and link building. Interestingly, those two things are very closely related to typical sorts of webspam techniques. Perhaps, it is time to focus SEO efforts on abiding by some standards, and make websites for people, not search engines.

EMDs are more than likely being targeted for violating keyword stuffing, other simple webspam techniques, and not building quality sites. The keyword phrase is in the exact match domain name, and this seems like an easy breezy technique. Start with the keyword from the Exact Match Domain and repeat it throughout the page.

The same is true for links. For a long time, the huge EMD bonus was that the website’s name is the money keyword one wants to rank for. It seemed OK to really overdo linking for the money keyword, as we thought Google couldn’t differentiate between that keyword being a brand name or not. Having lots of links and mentions in the body for your brand name (i.e., CEMPER) makes a lot of sense.

However, if you have a website www.BuyCheapSomething.com, it seemed that Google took Buy Cheap Something also as a brand name and ranked you fast for it.

Google Knew It, We Knew It

Don’t think for a minute that this EMD algorithm just came out of the blue. The patent for the EMD algorithm, Systems and methods for detecting commercial queries, was filed way back in September of 2003, and finally approved a year ago on October 25, 2011. Matt Cutts even talked about how they were going to change the EMD game in a video on March 7, 2011. Is Google being transparent by warning us that they are giving too much weight to EMDs?

We went through the weekly winners and losers list from SEOLytics to look for EMDs that had dropped sharply from rather stable rankings from the week before the update. We also did the opposite and found EMDs that actually gained in rankings after the update.

An Actual Example Of An EMD Loser

So, let’s just take a quick look at a few things a top loser did, and see if we can find if the site violated any of the guidelines.

We randomly picked one of the analyzed users and found this website:

www.businessliabilityinsurance.org

At first appearance, this site looks professional and has some reputable insurance company logos. Of course, we see the keyword phrases [business liability insurance] and [liability insurance] throughout the homepage. Perhaps, it isn’t overly done.

However, the FAQs page returns 38 matches for [business liability insurance] and 47 matches for [liability insurance]. Wow! The Guides page returns a total of 100 matches for [business liability insurance].

Are SEOs creating these types of text areas for users or search engines?

If you click thru those articles, you see that they are just spun content throughout the whole site, and are targeting every possible location with the keyword [Business Liability Insurance]. Not surprising that this site tanked — if you just look at the on-page factors of overdoing the commercial keyword.

What About The Link Profile?

Looking at the domain’s Power*Trust, we see a poor value of 6. More on that in a bit.

The anchor texts they used are striking when you look at this. There is not even an approach to mention the URL at substantial amounts (3% only, to be precise). The rest of the keywords are all money keywords in various combinations.

The Trust of those links are less than mediocre. There is only one business.com link at 5 and the remainders mostly below 3.

Other Trust Factors

An interesting Unable to connect error came up when trying to access the Get Quotes page. This error to the netquotes.com site could also be a factor. I would classify this as a mix between a sneaky redirect and cloaking. Perhaps, the netquotes.com site is not a trusted site. While doing our research, it did not appear to load.

Lastly, I really had to laugh at this deceptive maneuver shown below. As users, we value testimonials on websites.

When we see a photo of the person next to the testimonial, the value and perception that this company is trustworthy is even greater.

However, I know these people and they are not who they say they are. LOL! They are stock photos. This is not cool in my book!

Even the testimonials are keyword optimized.

What Is Trust & Why Is It Important?

Following the recent Penguin 3 update, I went back and reread the Another step to reward high-quality sites blog post on Tuesday, April 24, 2012. Again, I am highlighting this, because I believe it is important to read what Google tells us they are doing and not read a bunch of conspiracy and hate comments.

The fact is that Google wants highly-trusted sites in its rankings. And any attempt at manipulating the game will be dealt with. If you don’t believe me that Google is taking webspam seriously, here is a valuable quote from a paper that was written in 2005.

“Please note that according to our definition, all types of actions intended to boost ranking (either relevance, or importance, or both), without improving the true value of a page, are considered spamming.” (Web Spam Taxonomy by Zoltán Gyöngyi & Hector Garcia-Molina)

It is probably worth noting, Zoltán Gyöngyi is a research scientist at Google who went to Stanford and studied with Professor Hector Garcia-Molina, who was the principal investigator for the Stanford Digital Library Project, in which the Google search engine emerged.

We know from our research that by now Google is definitely targeting webspam and low-quality sites with its algorithms. So, what does it take to develop a high-quality site?

Following are some excerpts from our updated EMD Case Study, you can get the full report here.

Domains Compared By Power * Trust™

Since Google is always mentioning the overall quality of a site, I could think of no better metric than the CEMPER Power*Trust metric. (Admittedly, I am slightly biased here.)

Power means the strength based on the number and power of links (better than PageRank™). Trust indicates the implied Trust of the page in Google, according to a system similar to the Trust Rank patent. By combining both metrics, you can easily rate the overall quality of a domain.

The above chart is very clear; the average winner has double the amount of the Power*Trust compared to the losers. This results from a huge number of highly-trusted and very strong backlinks. While most of the losers’ backlinks are potentially low quality, it is pretty clear that the winners have way more links with high Power*Trust.

Our example, businessliabilityinsurance.org, has an even lower Power*Trust value 6 for their domain than the average loser with a Power*Trust value 8.

Domains Compared By Facebook Shares

Social media activity is (and should be) an ever increasing factor in rating the quality of a website. It’s a pretty clear factor to figure out if the audience likes the content and wants to share it with other people.

In general, a very popular website or brand automatically grows in social networks as soon as the reach higher rankings in Google.

This chart shows the huge gap between an average winner’s site with up to 180 shares, whereas the losers have only 17. That might result from the growing priority of a social network, as well as the fact that high quality websites in many verticals simply tend to attract more shares than low quality websites.

Domains Compared By Domain Popularity Growth

The Domain Popularity Growth (i.e., growth of new linking root domains per month) is one of the most important factors for a common natural link profile. A healthy website automatically grows over time because more and more webmasters tend to place a link to a higher quality site.

What we see here is that the winners have constantly more than double the linking domain growth per month than then losers. While the losers have built an average of 1239 new linking domains over the last 2 years (51 per month), the winners have built an average of 3193 new linking domains over the last 2 years (133 per month).

Is Google Really Evil?

Many SEO professionals and webmasters consider Google to be giant, ruthless regime. Many say they don’t follow their own Don’t do evil mantra when it comes to SEOs. Google is the Internet police, etc. etc… Like most regimes, Google constantly tells us that they are trying their best to be transparent. That is what the people want to hear. Google is a transparent corporation that aims to serve its people.

Surprisingly, most governments or corporations are transparent. You just need to read the correct information or listen in a different way. Large, hidden secrets are not implemented without first being written down.

Therefore, it should not be a surprise that the reduce ranking algorithms Google develops are sort of spelled out. Actually, they really are. Mainly, you just need to not violate the quality guidelines and build higher-quality websites.

Organize the world’s information and make it universally accessible and useful. That is Google’s mission.

Stop for a moment, and dream of a day when all search results on the first page are relevant to want you desire. Is it too much to ask for people to read and abide by some laws?

Google EMD Update – A Shock?

So, it comes as a bit of a shock when I hear outrage and panic about Google’s recent Exact-Match Domain algorithm. SEOs are angry, scared, and confused about how this update is harming their sites. Following the Sept. 29^th EMD update, it was amusing to read all the caps and cursing at Google in the blogs.

What did people expect? Learn a few basic SEO techniques, abuse the guidelines, and expect not to get caught?

Key Takeaways

• Review Google’s Quality Guidelines and make sure you are not in violation of any obvious tactics. For EMDs, keyword stuffing appears to be very common. Similar rules for normal brand domains apply. Many who got away with webspam techniques in the past, lost their free EMD lunch bonus.
• The overall domain strength is still a very important factor, which means the winners simply got higher quality links and more of them. This makes sense. The typical Quick EMD strategy didn’t need high quality links to rank very high.
• For all you social fans, it does appear that shares could provide a boost in rankings. Further research will need to be done to confirm this. However, I believe that Google takes the social signals and reads this a value of building a high-quality site. See Google bullet point, “Are the topics driven by genuine interest of readers of the site.”
• The domain popularity growth of the winners is so much higher and reflects the overall link and site quality.
• It’s not enough to just build links. More links and better quality lead to a better site, even if it’s an EMD. At least, after the long overdue EMD update.

Finally, Matt Cutts is really sticking to his words to crack down on Web spam in 2012. He closed a lot of loopholes, which surprised me, that were still working during the early part of 2012. I wonder if it’s his secret 30-day challenge to crack down on one old tactic per month?

Personally, I think there’s still a lot more to fix. Based on data we get from our Link Detox technology, which is part of our Link Research Tools software suite (independently reviewed here on SEL), there are still many sites with really bad link profiles and outdated SEO practices that are very easy to detect. I suppose that is also why we should expect these updates to continue on a regular basis.

Again, people, you should thank Matt and Google for letting you get away with EMDs for so long.