The only backlink analysis software you need. | LRT in Deutsch | Contact Us +43 720 883 736(619) 832-0736+441443606363(204) 808-0736
Home > Blog > Case Studies

12 Lessons Learned in Penguin 2.0 Analysis

12 Lessons we’ve learned from Google’s Penguin 2.0

12 Lessons learned from Google’s Penguin 2.0. Analysis

Penguin 2.1 launched on the 4th of October and we’re working on our Penguin Analysis just as we promised. But for now we have compiled twelve “lessons learned” from the case studies we have already published on Penguin 2.0.

This analysis was created by our Certified LRT Professionals Geraldine and Guntram creator of previous case-studies too, and I look forward to meeting them in our private CLRTP meeting today as well and discuss things like Penguin 2.1, the War on SEO, our Link Detox Genesis™ release candidate and many other things.

We summarized findings and observations which I find very interesting. We tried to keep any kind of statistics out of this meta-cases study as well, because the sample size is too small, but find our qualitative observations very interesting, so we hope you like it too.

Thanks Geraldine & Guntram!

- Enjoy & Learn!
... and please help spread the word.

Christoph C. Cemper

PS: Over the past couple months a few people got the impression that our case-studies on single websites should serve as “global proof” for any kind of assumption or observation we drew when manually reviewing a site. That’s not the case.

In our case studies we inspect and evaluate a web site, just like you would do in a site audit, in a qualitative way (by the SEO) the findings, comment on it and help the reader follow the process.

A sample size of N=1 or N=12 doesn’t make good statistics, too few data points i.e. websites to draw global conclusions from it. If you think we did statistics in our case-studies you should go re-read them.

If you want to understand why some people prefer the word “Filter” in relation to Penguin vs. “Penalty” you should read Derek Devlin’s (another CLRTP btw) excellent introductionary post on Google Penguin.

Table of Contents

LRT Penguin 2.0 Case studies

This analysis was done by Geraldine Edel (GE) and Guntram Bechtold (GB), with the author marked at the final comment for each section.

The findings from the single case-studies listed below were aggregated to answer a simple question - to we feel there are common pattners?


1. Was Anchor Text overdone?

Site Analysis Result No data - Yes. had 66% money keywords compared to 29% of the total average! YES Yes. had far more money keywords than its competitors: 82% to only 4%! YES Yes.'s money keyword ratio was the highest among its competitors. Further, its keyword distribution seemed very unnatural. YES Yes. The website's money keyword ratio was at 56% compared to 9% of the TOP3. Further, there have been many .edu backlinks with heavily skewed money keyword anchor text distribution. YES Maybe. The website had a money keyword ratio of 31%. It had more or less an equal percentage of links allocated to brand, compound and money keywords. - Yes, the website had a very high money keyword ratio of 75%. The total average of its competitors only lied by 24%! YES No. The website had some money keyword anchor texts, but not in significant volume NO Yes.’s had a money keyword ratio of 84%! The total average is only 39%. YES No. Although the website’s keyword ratio was a bit unbalanced, its money keyword ratio was almost half of the competitor’s money keyword ratio. NO No data - No data -
Result: 6 out of 8 had too many exact match anchor texts 6/8

Penguin 2.0 definitely looked at the anchor text ratios. We clearly validated this fact on 7 out of 9 case studies. That fact is obvious and clearly measureable for sites. Even more interesting is that sites with just a couple of links, let's say up to 100 total backlinks, appeared to be affected. Generally it needs to be made clear that every niche got slightly different rules and tolerances. (GB)

2. Did the websites have more LRT Power than LRT Trust?

Site Analysis Result Yes. Significantly. Power was at 6 and Trust was at 3. - Yes. Significantly. Power was at 6 and Trust was at 4. YES Yes. Significantly. Power was at 5 and Trist was at 3. YES No. Power was at 5 and Trust was at 6. NO Yes. Significantly. Power was at 6 and Trust was at 4. YES No. Power was at 4 and Trust at 4. NO Yes. Significantly. Power was at 3 and Trust at 1. YES Yes. Significantly. Power was at 8 and Trust at 6. YES Yes. Significantly. Power was at 5 and Trust at 3. YES No. Power was at 5 and Trust was at 6. NO Yes. Significantly. Power was at 5 and Trust at 3. YES Yes. Massive. Power was at 5 and Trust at only 2. YES
Result: 9 out of 12 websites had more LRT Power than LRT Trust or vice versa. 9/12

Power*Trust is a very powerful indicator. It visualizes the vulnerability towards getting banned by Google for a specific website: If power and trust are equal, we can see that complex relationship of the site towards other sites and authorities is okay. Some websites got significantly higher power than trust, the signal for Google is obvious: These websites do not have generic growth but have been developed by using steroids.

Why is a big gap between Power and Trust bad?

Typically, webmasters try to build strong links from the start in order to fuel their own PageRank. Getting strong links is hard and getting them from low trust sites (i.e. the spammy ones that sell links, etc.) is easier. For instance, got tons of .edu links, yet these .edu links had low power and practically no trust. That's why the website had a LRT Power of 6 but a LRT Trust of only 4. (GB)

3. Did they have too many sitewide links?

Site Analysis Result Yes, a high ratio, 990 Class C IPs generated more than 117.000 Links - a high number of sitewide links. YES No. All of competitors had a very high ratio of sitewide links. had a comparatively small sitewide ratio. NO No, and its competitors all had an equally low sitewide links ratio. NO Yes, had a very high ratio of sitewide links YES No data - No data - Yes. The redirected domain had 14.819 sitewide links out of 14.904 total links! YES Yes, many sitewide links. 5 to 10 times more than the competitors. YES No data - Yes, compared to its competitors, had the second highest site wide ratio! YES No data YES No. Sitewide links haven’t been an issue NO
Result: 5 out of 8 websites had significant sitewide links. 5/8

6 out of 9 websites had excessive sitewide links. Sitewide links are not beneficial for the growth of a websites's ranking. On the contrary, sitewide links can harm a website. for instance had excessive sitewide links through its 301 redirects. created some problems using sitewide links since around 990 Clas s C IPs sent more than 117.000 links. It is pretty unrealistic that one could motivate 990 webmasters to link to thousands of times for nothing in exchange

4. Did the websites have paid links?

Site Analysis Result Yes, a couple of obvious paid links with exact match anchor texts could be identified YES No, there haven’t been any paid links. NO No, there haven’t been any paid links. NO Yes, there have been high quality paid links YES No, there haven’t been any paid links. NO Yes, it appears that there have been paid links. YES Yes, with more than 50% of paid links (very obvious ones) from YES No, there haven’t been any paid links. NO No, there haven’t been any paid links. NO Yes, there have been many paid links YES No. NO Yes, bad SEO directories and paid blogposts YES
Result: 6 out of 12 websites had paid links. 6/12

Identifying high quality paid links isn't an easy thing. The upcoming Link Detox Genesis™ tries to place a lot more emphasis on finding those better made paid links, so we are looking forward to the results we can get by then.

Having links from paid blog articles – like in the case of - will get you banned as well, but not only because these links are paid ones, but also because mostly these websites are participating in link schemes. These link farms got no content focus, typically link out to everyone paying the fee. Using sites like these are just dangerous and will put you into the deadly danger zone.

5. Was the websites Power*Trust distribution unnatural?

Site Analysis Result Yes, the percentage of LOW quality links with Power*Trust 0 and 1 is WAY higher than the average for the TOP10, TOP5 and especially the TOP3. The percentage links with higher Power*Trust is really small compared to the competitors. YES Yes, had about 2 times more strong Power*Trust links, compared to other websites from this niche. On the other hand, far less weak links (Power*Trust < 2). YES Yes, the website had a lot of Power*Trust 1 and 2 links. The percentage of links from websites with higher Power*Trust is much lower: only 3% of’s links had Power*Trust 5 – 7, compared to 20 % of its competitors. 2% of’s links had Power*Trust 8 – 12, in comparison,’s competitors had about 25% Power*Trust 8 – 12 links. YES No. Power*Trust was fine NO Yes, Power*Trust was very low. 2x more weak links than comparable sites. YES Yes, Power*Trust was very low. 93% of links have been equal or below 1. YES Yes, the websites had too many weak links with Power*Trust 0. YES Yes, the website’s links had a significant lower Power*Trust than its competitors. YES Yes. The Power*Trust of’s links was significant lower than for competing sites. Further, the website had a high proportion of LRT Trust 0 links and less LRT Trust 2, 3-4, 5-7 or 8-12’s YES No.’s Power*Trust ratios fitted quite well into the link profile of its competitors NO Yes. had many more weak links than its competitors. YES Yes. had many more weak links than its competitors. YES
Result: 11/12 had unnatural Link Power*Trust 11/12

Several websites had really toxic link graphs. They have been outside of the normal variance and had 2 times more weak or strong links than similar sites in the niche. One of the toughest cases was More than 93% of the website's links had a Power*Trust below 1. Almost unbelievable that Zombie-Sites like these ever ranked. (GB)

6. Have sneaky redirects been involved?

Site Analysis Result Yes, had lots of unhealthy and unnatural looking redirects. Although the redirected domains have been on-topic ones, the links themselves were merely paid ones. YES No. No sneaky redirects have been found NO No. No sneaky redirects have been found NO No data - No data - No data - Yes, had 1,6% redirects. All of them were totally off-topic ones! YES No data - No. Although many redirects have been in place (5,8%!), most of them came from variations of Icelolly brand name, including other TLD extensions as well as possible misspellings. NO Yes, had at least 279 sneaky redirects from the website YES No, although there have been found 153 links that have been redirected, in the case of this amount of links is very small compared to the total amount of backlinks of more than 90.000. NO No data -
Result: In 3 out of 7 cases sneaky redirects have been involved. 3/7

Redirecting expired domains that have lots of backlinks to transmit link juice to a specific domain and therefore achieve high rankings has been a loophole for several years. But as we all know, it is just a matter of time until Google closes those kinds of loopholes.

As could be seen in previous case studies (e.g. see case study, redirecting expired domains has been a popular linkbuilding tactic. At the latest after the rollout of Penguin 2.0, redirecting expired domains that have lots of spammy and/or off-topic links doesn't help anymore to achieve better rankings, on the contrary, these actions can be a reason for getting hit by Penguin 2.0 or for getting a manual penalty by Google.

As has been identified in some case studies like, sneaky redirects hurt websites link profiles and have been one of the reasons for the massive drop in SEO visibility. Redirecting domains is an important issue in SEO. There are some points you have to keep in mind:

  • Redirects can hurt a website if the redirected domain is not on-topic and has got an unnatural link profile.
  • Redirecting expired domains can still be a useful way to get better rankings, but they have to be 100% on-topic! For instance, redirected a totally off-topic page, including anchor texts like “Value my car” or “how much is my car worth”. Redirecting lots of backlinks with off-topic anchor texts does affect your link profile in a negative way!
  • Further, redirects transmit good link juice as well as bad ones! Therefore it is necessary to check each link before redirecting it, otherwise this tactic can harm your website a lot if the amount of bad links is high compared to the websites total amount of backlinks. For instance, in the case study for 153 links have been found that are redirected to For this amount of links is very small compared to the total amount of backlinks of more than 90.000. But for websites with far less backlinks this amount of redirected links could have a big impact on the main domain's link profile.
  • Redirecting domains can also be a good possibility to hide some parts of its own link profile. Contrariwise, if you want to check your competitors link profiles, watch out for 301 redirects! Also keep an eye on your own 301 redirects! Check all redirects to your website with which you are not familiar. This redirects can be a consequence of a negative SEO attack and could have a high amount of links with anchor texts you don't want to be associate with!
  • So, does Penguin 2.0 punish sneaky redirects? Well, according to the LRT case studies, 3 out of 7 losers had sneaky redirects. If we can expose sneaky redirects, Google can do that for sure as well, so I suggest yes, Penguin 2.0. punishes sneaky redirects. (GE)

7. Has unnatural link growth been involved?

Site Analysis Result No data - No, the website had a good and healthy link growth NO Yes, had proportionally a higher link growth than its competitors. YES No data - No data - No data - No data - No data - No. After got hit by Penguin 2.0 its linkbuilding has almost stopped. This is of course an unnatural pattern, but that happened AFTER the Penguin 2.0. rollout, before its link velocity was O.K NO No, no unnatural link velocity could be identified NO No, its link velocity was quite balanced NO No, there was no unnatural link velocity involved NO
Result: In 1 out of 7 cases unnatural link velocity has been involved 1/7

Unnatural link velocity can be very harmful for every website. What I've seen quite often were websites that have been online for ages and suddenly started doing lots of linkbuilding. The consequence, the website's link growth exploded from 0 to 100 – in Google's eyes of course a very unnatural issue and therefore most likely an indicator for manipulation. Of course, if it is a big brand launching a new website doing some advertising on TV there would probably also be a hype like that, but there would also be some signals around it some social media buzz that makes it natural for Google.

Otherwise, stopping the whole link building process from one day to another can also be very harmful. Again, this is very unlikely and therefore another signal for manipulation.

As could be seen in the case studies, only 1 out of 7 websites had an unnatural link velocity. Nevertheless, always watch out to have a similar link growth as your competitors! (GE)

8. Have link networks been involved?

Site Analysis Result Yes, thousands of links came from only 990 different Class-C network and 1.756 domains. That’s about 2 domains per Class-C on average and especially 121 links per domain on average. YES Yes, there have been 1000s of websites from article link networks YES No data - Yes, many of links have been from the same network, many have been registered by the same person, others have been hidden behind different registrars, even more hidden behind private registrars. YES Yes, 112 links have been from the same IP and all had a LRT Trust of 0 YES Yes, almost 20% or 289 of all links came from one single IP YES Yes, although it was suggested that being part of link networks is not the reason for the de-indexination in the case of, there have been found lots of links from cheap-labor “offshore SEO shops” YES Yes, 3.9% of its links have been from the same IP and from two others at 2%. Further, 4,2% links were coming from the same Class-C IP YES No, link networks have not been used NO Yes, tried to address different target groups with different websites (e.g. “” is especially for women, “” addresses people who need a travel insurance e.g…), but it is very likely that this strategy was seen by Google as participation in linkbuilding networks YES Yes, was using link networks like web directory networks or business directory networks YES No, no links from the same IP have been originating NO
Result: In 9 out of 11 cases link networks were involved 9/11

Links are essential to gain good rankings. We all know that. But getting a bunch of links by using a link network is the worst thing you can do. But what is a link network? Well, to keep it simple: a group of websites that are connected via same IPs, ClassCs, registrars, and so on. Especially if your links are from dubious companies that offer lots of links for a very low rate you can be sure that you bought links from a link network. Alright, that's obvious, I know. To identify, if you get/bought/exchanged links from a link network is usually not that easy.

Nevertheless, we all know that Google's Webmaster Guidelines restrict the usage of link networks. They warn against the use of link schemes and specifically mention “using automated programs or services to create links to your site.”

So did Penguin 2.0. punish the usage of link networks? Well, in 9 out of 11 cases link networks have been involved, so I would say yes. Again, if we can identify links from link networks Google can do that as well. Keep in mind: Links from link networks are very easy to get, so of course Google tries its best to prevent those practices. (GE)

9. Did any unnatural country ratios occur?

Site Analysis Result No data - Yes, didn’t have a lot of non-US or non-EU links. But seemed to be more active in Europe than its competitors. Only 56% of its links have been from US based sites, although’s business focus lies on the US market. YES No data - No data. - Yes, only 38% of’s .edu links have been from US-hosted websites YES Yes, 23,7% of its links came from Polish websites although’s content is in English YES Yes, wants to rank in UK but most of its backlinks have been from US websites. Also this country distribution looks very unnatural! YES No data - Yes, had overdone it with US links YES No, had less US links than its competitors but as the aim of the website is to rank in UK, this ratio is fine. The ratios for the other countries fitted perfectly into the link profile of the competitors NO No data - No, the geographic location of’s links has been balanced: Mainly US, but also some other countries NO
Result: 5 out of 7 had an unnatural country ratio 5/7

From which country your links are coming from DOES matter, of course. As could be seen in the case of, more than 20% of its links came from Polish websites. It is normal, natural and inevitable to get links from different countries, but the majority of your links have to be from your main target market. If this is the Polish one or the US one depends on your business model of course, but don't be surprised if you don't achieve good rankings in Google.US while having most of your links from Asian, Russian or European websites.

Further, links from countries that don't really have anything to do with your business/website are mostly spammy ones, generated automatically by supposed “SEO agencies”.

But also for country ratios apply the same rule as for other parts of a website's link profile: Always try to have similar ratios as your competitors to be on the safe side 😉 (GE)

10. Has there been an unnatural Deep Link Ratio?

Site Analysis Result No data - No data - Yes, had a higher deep link ratio (59%) than its competitors (49%) YES Yes, had a higher deep link ratio than its competitors. Even though 5% is not that much, it is still unnatural YES No data - Yes, while competitors had an average 98% to 2% deeplink to startpage link ratio, ratio looked exactly the opposite: 12% to 88% deeplink to startpage link ratio YES No data - Yes, its deeplink ratio was a bit unbalanced YES No, it fitted fine NO Yes, had 13% more startpage links than the total average YES No data NO Yes, it had a bit too much startpage links YES
Result: 6 out of 8 had an unnatural deep link ratio 6/8

So what's the perfect deep link ratio? 50% - 50%? 30% - 20%? The answer is yes, no and maybe. We all agree in the fact that a website's link profile has to seem natural in order to avoid any punishment by Google. What is natural? Well, look at your competitors and you'll find the answer 😉 (GE)

11. Did the websites have an unnatural Link Status Ratio?

Site Analysis Result No data - No data - Yes, had 96% follow links compared to 88% follow links of its competitors YES No. had the highest percentage of follow links. But as the gap is very low, namely 93% follow links compared to 91% of its competitors, I would say that its link status profile is still balanced NO No data - Yes. 100% of’s links are follow ones! YES No data - Yes. has much more NOFOLLOW links comparing to competitors YES No. Its link status profile was balanced NO Yes. It had 82% Follow links compared to 75% of the TOP3 Average YES Yes. 27% nofollow links compared to 18% total average YES No data -
Result: 5 out of 7 had an unnatural link status profile 5/7

We all know, follow links pass link juice, nofollow links do not. Nevertheless, a healthy link profile needs both. But what about having more Nofollow links than the competitors? Does it even look more natural, as these links do not transmit link juice and therefore do not effect rankings? Well, that's a good question. I'd say no, it also looks suspicious. This can might be seen as a failed try to get links out of blog comments or something like that, without recognition that they are nofollow. Together with other unnatural patterns, too many nofollow links can be harmful as well in my opinion. But as I said, this question is very hard to answer.

So what ratios are the best link status ratios? Again, check your competitors 😉 (GE)

12. Did the websites have a high Link DTOX Risk?

Site Analysis Result Yes, only had 43.8% healthy links YES Yes, has links from 148 pages that are not indexed anymore. All in all its link profile has been quite risky YES No.’s link risk was moderate NO No. Its link risk has been low. NO Yes, only nearly 20% of its links has been healthy YES Yes, had a high link risk YES Yes, the website had a deadly link risk! YES No, the website had moderate link risk NO Yes, it had a very high link risk. 76% of all links have been either suspicious or toxic ones YES Yes, had a high link risk YES No,’s link risk was almost as high as its competitors link risk NO Yes, had a deadly link risk. The total of Toxic and Suspicious links was at 87%! YES
Result: 8 out of 12 link profiles have been risky 8/12

13. Results Analysis

1. Was Anchor Text overdone? 11/12 overdid anchor texts
2. Did the websites have more power than trust? 9/12 had bad Power*Trust Ratio
3. Did they have too many sitewide links? 9/12 had sitewide links
4. Did they have paid links? 7/12 used paid links
5. Was the link’s Power*Trust distribution unnatural? 11/12 got unnatural link P*T
6. Have sneaky redirects been involved? 3/7 had sneaky redirects
7. Has unnatural link growth been involved? 1/7 had an unnatural link growth
8. Have link networks been involved? 9/11 had links from link networks
9. Did any unnatural country ratios occur? 5/7 had an unnatural country ratio profile
10. Has there been an unnatural Deep Link ratio? 6/8 had an unnatural deep link ratio
11. Did the websites have an unnatural Link Status Ratio? 5/7 had an unnatural link status ratio
12. Did the websites have a high Link DTOX Risk? 8/12 had a toxic link profile

So, what was Penguin 2.0. all about? What lessons have we learned from Penguin 2.0?

Well, getting hit by Penguin 2.0. can have several reasons. As could be shown in this “Meta Case Study”, high exact match ratios as well as toxic links are a problem for most of the analyzed Penguin 2.0. losers. Sitewide links of course did also appear in most of the cases, but first avoiding sitewide links is a big challenge, second, it really depends what anchor text the sitewide links have got – if it is a branded term, I would not worry too much about it – if it is a hard anchor text keyword, be careful! Weak links and spammy links, it doesn't matter if they come from sneaky redirects or obviously link networks are for sure issues that can lead to massive drops in search visibility. What we've learned so far is that Google is becoming better and better in exposing unwanted SEO practice. Do you want to achieve better rankings? Build links because links are essential for that aim, but try to that according to Google's guidelines – at least as much as possible.

In this spirit: standard “Black Hat SEO” Adé – “White Hat SEO” Olé! 😉

Thanks, Geraldine and Guntram

Certified LRT Professional
This case study was written by Geraldine Edel and Guntram Bechtold, both Certified LRT Professionals

A word from Christoph C. Cemper

Guntram and Geraldine already showed proficiency in doing a backlink profile and SEO audit using the LinkResearchTools in their original case-studies and helped summarize findings from all published Google Penguin 2.0 case studies. A great series finale for the Penguin 2.0 case studies, and we're getting ready for the 2.1 case studies now.

The goal of our LinkResearchTools certification program is to provide our user community and clients a high quality service, and our certified experts are key to that.

I look forward to future work and personally recommend Geraldine Edel and Guntram Bechtold to work with you whenever you get a chance!

Certified LRT Professional Guntram Bechtold

Certified LRT Professional Guntram Bechtold


Christoph C. Cemper
Christoph C. Cemper is the CEO and Founder of LinkResearchTools and Link Detox. A well-known and distinguished expert in SEO who started link building for clients in 2003, building the Link Research Tools since 2006 and marketing it as SaaS product since 2009. When the famous Google Penguin update changed the rules of SEO in 2012, Christoph started Link Detox, software for finding links that pose a risk in a website’s backlink profile. He introduced ongoing link audits and risk management to the market in early 2011. In 2015, Christoph introduced Impactana, a new technology platform and SaaS product to measure the success of content beyond "social buzz", to find content, videos and people that make an impact.
Christoph C. Cemper
Christoph C. Cemper
Christoph C. Cemper

Latest posts by Christoph C. Cemper (see all)


  1. Subhash Jain on October 21, 2013 at 1:40 pm

    I liked this case study. SEO is a changing world every moment.

  2. Geraldine on October 22, 2013 at 4:51 pm

    Thanks a lot Ruben and Subhash! 🙂

  3. Byron Hardie on November 12, 2013 at 5:30 am

    Thanks for putting this together. While we know that correlation does not equal causation it could provide some directional data that others can test and use in their own evaluation. While the Penguin updates did focus a lot on links, it was really targeting web spam in general (which many people seem to forget unfortunately) so it would be interesting to see if there were other on-page correlations or over-optimization issues with the sites in the case study.

    Thanks again Chris. Nice to see other search scientists out there getting neck deep in the data!

Leave a Comment