Case Study for debtconsolidationcare.com
We are proud to present the 5th deep dive case study by a Certified LinkResearchTools Professional. We greatly appreciate you sharing this quality piece of work by Michael Marshall.
Christoph C. Cemper
Looking into this Penguin 2.0 drop in visibility
This case study is set out to analyze the reasons for the drop in visibility, that debtconsolidationcare.com experienced after the global rollout of Penguin 2.0 update.
We will look at their website like any professional SEO would conduct a SEO and link audit. This research will look into Spammy Links, Suspicious EDU links, Low Quality Links, Link networks, Bad linking neighborhoods and much more.
- Quick Domain Compare – how do they match up?
- Low CEMPER Trust™
- Suspicious EDU Links
- Anchor Text
- Host Country
- Spammy Content
- Quick Competitive Landscape Analysis with CLA
- Anchor Text
- CEMPER Trust™
- De-indexed Sites
- Malware or Malicious sites
- Hacking or Pornography sites
- Link networks and Bad Neighborhoods
- Digging Deeper
- Latent Competitive Analysis: LCA based on CLA
- A little help from Artificial Intelligence
Looking at the visibility, we can see a massive drop in traffic right around the time Penguin 2.0 rolled out, there was a steady decline in SEO visibility from December of 2012 onwards, but the drop in mid May 2013 was massive and not comparable to the gradual decline throughout the year.
A first quick look at the affected debtconsolidationcare.com domain and some of their competitors
Using the QDC tool, we compare them on some important metrics
And the result is below:
Quick Domain Compare – how do they match up?
Low CEMPER Trust™
When you look at debtconsolidationcare.com compared to these competitors it is definitely deficient in the area of CEMPER Power*Trust™, and especially in the area of CEMPER Trust™.
Suspicious EDU Links
Another thing that stood out to me was the high number of edu links relative to the competitors excepting www.wellsfargo.com.
By using a filter slice on the list of links, I was able to dig deeper into that number, 692, and I saw that the anchor text distribution for those EDU links consists almost entirely of money terms!
It also became clear that all of those edu links are on pages that have low or no CEMPER Trust™, in fact 99% edu links had a CEMPER Trust™ of 0 and the highest of any is 1
I also noticed that the majority of those edu links are from non-US countries. Only 38% of the EDU links are on sites hosted in the US.
So I decided to examine some of the pages. Below is an example of some of what you’ll find.
One of those links is http://ce.sharif.edu/~a_ghasemi/bbclone/doc/rss/genevapaydayloan.html
First line of text reads like this:
“Many borrowers find the higher payment out of reach and choose a 30-year mortgage. Justice scalia, or that prompted his question in any event, is.”
The remainder of the content on the page is even worse, clearly spammy, low quality content, like a mosaic of machine generated or scraped content with bad grammar and semantic coherence that is dodgy at best.
Quick Competitive Landscape Analysis with CLA
Here you can see that for debtconsolidationcare.com the distribution of anchor text is slanted much too heavily on money terms, not just with edu links as shown above but with the link profile as a whole.
Once again, the level of CEMPER Trust™ is very low and a huge deficiency for debtconsolidationcare.com compared to competitors just as it was shown for the edu links earlier.
Link Detox Risk Analysis
Only 19.8% Healthy links is an extremely toxic backlink profile.
Quite a number of them pose a huge risk, over 10%!
Those are links that especially trip multiple Link Detox rules, which can be reviewed on a per link basis later.
Here is an example of one with deadly risk that actually trips multiple Link Detox rules:
Click the (+) and you can see more detail showing which rules were tripped, in this instance: Tox1, Susp5, Susp1, and Susp2.
Other very disturbing finds are as follows:
426 have been de-indexed by Google. This is usually a sign for a link network penalty or at the very least very weak sites, duplicate content or any other site that at least doesn’t add any link value.
Malware or Malicious sites
504 are listed as Malware, Malicious, or Virus
Hacking or Pornography sites
275 are listed as Suspicious (Hacking, Suspicious or Pornography)
Link networks and Bad Neighborhoods
112 links are all on the same IP and all have CEMPER Trust™ of 0; a clear sign of a possible link network. That IP address in question is: 188.8.131.52 and there are 157 domains hosted on that IP.
a. There is also an interesting warning from one reverse IP lookup service:
“It appears that the web server located at 184.108.40.206 may be hosting one or more web sites with explicit content. The web sites in question are highlighted in red below. There is a possibility that all of the web sites on this web server may be blocked by web filtering software. Search engine rankings for these web sites may be affected as well.”
This clearly puts a site into a Bad Neighborhood!
Summary and Conclusion
1. Domain Comparison
- The site is very deficient in the area of CEMPER Power*Trust™, and especially in the area of CEMPER Trust™ when compared to competitors.
2. EDU Links
- EDU links consists almost entirely of money terms which makes for a heavily skewed (in the wrong direction) anchor text distribution.
- A whopping 99% of edu links had a CEMPER Trust™ of 0
- Only 38% of the EDU links are on sites hosted in the US.
- EDU Links are spammy and have very low quality content that would never pass something like a Panda review.
3. Competitive Landscape
- The anchor text distribution for the site is slanted much too heavily on money terms.
- The level of CEMPER Trust™ is very low when compared to competitors.
4. Risk Level Assessment
- The link profile has a large number of links on sites de-indexed by Google
- The link profile has a large number of links on sites with risk of Malware, etc.
- The link profile has a large number of links on sites with risk of Hacking, etc.
- The link profile has over 100 links from the same IP address, most likely a link network, that it is in a bad neighborhood
All of this adds up to a backlink profile that should not stand a chance of surviving Penguin 2.0.
The nature of the EDU links suggests a linking scheme across sites hosted in multiple countries intended to boost the authority and trust of the target site simply from the fact that they are EDU links with no concern for the quality of the content on those linking pages nor for the trust level or authority level of those linking pages and sites themselves.
The anchor text distribution is too heavily slanted toward money terms. Also the trust level is too low when compared to the competitive landscape. This does not result in a backlink profile that looks natural nor is it strong enough to compete.
There are too many links in the profile that are of high risk levels in many different ways, clearly putting the site in a very bad neighborhood.
The only solution here is to remove the risky and suspicious links and work on building new links that have higher trust levels. Work also needs to be put into getting more links using branded or mixed anchor text to get a more natural looking profile.
Often after an update by the search engines, people panic and are at a loss as to what they should do. It’s actually at times like this that it is most clear what you should do. You learn from the winners! You also learn from the losers. You just have to know how to study them. You need enough data. There is an enormous amount of information provided by LinkResearchTools and that is why analyses based on it can be so informative.
I believe the data LRT offers is very powerful in answering the question, “What just happened?” However, it is just as powerful in answering the question, “What do I do next?” . . . if you know how to look at that data.
Latent Competitive Analysis: LCA based on CLA
With all this data available, I thought it could prove useful to conduct a deeper analysis of that data. What if we mined it, you know, data mining? You’ve probably heard of latent semantic analysis (LSA) where you discover hidden or latent semantic relationships between words in a corpus of documents. I thought it would be interesting to do something similar with the data from the competitive landscape analysis (CLA) in LinkResearchTools. Hey, with that much data, how could you resist? We’ll call it latent competitive analysis (LCA), hidden or latent competitive relationships between backlinks of competitors in a landscape.
A little help from Artificial Intelligence
So, we can throw a little artificial intelligence at it and see what we can learn in more detail from the CLA data about what should be done moving forward. To conduct our LCA, I used an eigenvector-based multivariate analyses technique known as principal component analysis, (PCA). It’s not as complicated as it sounds. Okay it is; but there are plenty of tools that can do it for you. For example, the Excel Add-In XLSTAT is what I used.
The output from CLA has to be converted into the proper form to use as input for the LCA, so I wrote a program in PHP to do that so I wouldn’t have to do it manually. Hey, wouldn’t you? This transforms the CLA output of 21 selected factors into a sparse data set with 755 related factors, custom designed by this landscape. Once I finished the LCA, here’s what I found as the top 12 things for this landscape:
The most important things to be mindful of because of competitor performance in this landscape are:
- Links from sites with a Site Theme of Reference
- Google Plus activity around the site
- CEMPER Trust™ for the domain of between 5 and 7
- TitleRank Home of 1
- FB Shares
- Links from sites with a Site Theme of Business/Economy
- Proper proportion of FOLLOW links
- CEMPER Power™ for the domain >= 4
- Healthy proportion of IMAGE links
- Healthy distribution of Anchor text with Keyword Classification of Compound
- Healthy distribution of Anchor text with Keyword Classification of Brand
So as you peruse the LRT data, and as you go about link building, these would be factors which you’d give special attention to for this landscape. That is the specificity made possible by conducting LCA on the CLA data and this is what I would add to my initial set of recommendations.
Appendix: The Geeky Stuff
The following information is not included to provide any additional insight regarding the case study itself or to the recommendations already given. It serves merely as a little peek behind the curtain showing what the PCA process looks like. The scree plot of the PCA component factors is below:
F1, F2, F3, etc. correspond to those latent competitive relationships in this landscape.
The plot below looks at the 755 metrics and plots them against their correlation with the 1st (F1) and 2nd (F2) principal components. The red box highlights those factors with the highest scores.
Remember this is only plotted against the first two principal components. The LCA based recommendations given above were based on 5 principal components but I can’t easily plot that for you. Here is a list of some of the factors in that red box.
- Links from sites with Site Theme Business/Economy
- CEMPER Trust™ for the linking domain of 3
- FB Shares up to 151
- Image Links
- Links from sites with Site Theme Charitable Organizations
- CEMPER Trust™ for the linking domain of 4
- FB Likes up to 658
- Links from sites with Site Theme Education
- Links from sites with Site Theme Art/Culture
- CEMPER Trust™ for the linking page of 3
- CEMPER Power * Trust™ for the linking page of 6
- CEMPER Power ™for the linking page of 3
- CEMPER Power™ for the linking domain 4
- Links from sites with Site Theme Reference
FYI - I plan to host a Google hangout walking through this study with Q&A at my Google+ Community – SEO Mastermind:
This case study was written by Michael Marshall,CLRTP and was reviewed and approved by Christoph C. Cemper for publishing as Certification work for the Certified LRT Professional level.
A word from Christoph C. Cemper
This analysis was conducted and post written by our LRT Associate Michael Marshall, CLRTP.
Michael not only condensed the major issues down to crisp results but also suggest completely new methods to look at the problems, something we consider for implementation in LinkResearchTools as well. Therefore I'm happy to certify Michael for the Certfied LRT Professional level by approving and publishing his research on our site.
This is Michael's next step towards the Certified LRT Xpert level which is pre-requisite for the Certified LRT Agency certification. Both will qualify him to receive consulting leads from us. Our goal is to provide our community and clients a high quality service, and our certified experts are key to that.
I am sure Michael will continue to perform as well in our program and for clients. As you can see from his work already, I can really recommend him to work with you whenever you get a chance!