1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Why Google discontinued PageRank and why DA/PA or TF/CF is not very accurate

Discussion in 'White Hat SEO' started by tar729, Jul 5, 2016.

  1. tar729

    tar729 Registered Member

    Joined:
    Jun 27, 2016
    Messages:
    90
    Likes Received:
    108
    Some of you might have seen my previous threads here and here where I attempted to talk about my ideas using Google translator.

    I have had one of my articles translated so hopefully it will be easier to understand what I’m talking about. Do ask me questions but keep in mind I need a translator to reply so I might not be as fast as I would like.


    About Me:

    I have been working in SEO industry in Russia since 2009 and my speciality is in understanding the foundation of SE algorithms, usability and behavioural patterns in Yandex and Google. I have developed a system that analyses SE ranking factors, semantic relevance and measures the weight of links. At the moment this system is tuned for Russian market but parts of it can be used in English speaking countries as at its core the algorithms are the same.


    Glossary:

    Before we get into this subject I feel that some terms need to be explained because they might be new to some people. Below is the small glossary of terms to help you understand what I’m talking about:


    · Donor = a page linking out to another page

    · Receiver = a page the donor is linking out to

    · Semantic relevance = topical relevance i.e if donor page is talking about dog food and receiver is about food for small dogs those pages with have high semantic relevance

    · Semantic core = collection of key terms with close logical connection to the main search term, arranged by logical relevancy and not by search volume


    N.B. Semantic core is a very complex subject. I will put together another article where I will explain this concept in great detail.


    Why PR is dead?

    Google said that it discontinued PageRank because “PR no longer reflects the algorithm accurately”. That is true but there is a lot more to it.

    PR, as a metric, does not reflect the overall quality of a page but only indicates the quality based on incoming links and the traffic the page generates. Of course, the formula became more complex. Over time, Google attempted to take into account not only the quantity of incoming links but also their quality, by analysing the pages where the links came from. Nevertheless, when analysing the quality of pages and outgoing links they never considered semantic relevance of the donor paged to the receiver or anchor text.

    This approach was flawed and became of this started to affect the quality of SERP results. Web masters would concentrate on PR alone when selecting possible link building targets, mistakenly thinking that the higher the PageRank, the more powerful the link from that page.


    What is link relevancy?

    Even now very few people within SEO community assess the quality of the link based on how relevant the anchor is to the content on the donor page. For most this is still a secondary factor at best. Of course, every SEO specialist does his best to acquire a relevant link but the human view of the relevancy is very different from algorithmic view.

    One of the ways to test the relevancy of the page to a specific key phrase is to run a “site” operator in google. For example, you would type site:yourdomain.com dog food to see if there are any page on your site relevant to keyword “dog food”. Google will return result in the order of most relevant (in their opinion). If your landing page for “dog food” is no. 1, it means Google sees it as the most relevant on yourdomain.com. If on the other hand, your target page is buried within other pages on yourdomain.com, that means that Google does not see your target page as the most relevant. Building a link with anchor “dog food” will not have the maximum effect and might even be considered as spam.

    Eventually, it became clear that PageRank as a metric became detrimental to the entire search engine because it misled SEO specialists and encouraged non-relevant linking between sites. This, in turn affected the quality of SERPs and the user experience. As a result, the decision was made to remove PageRank as a factor in decision making process when considering your link building strategy. However…


    DA/PA or TF/CF?

    Due to the secretive nature of Google and other search engines and the lack of understanding of fundamentals of search algorithm by the public, many myths and legends were created within SEO industry.

    Removal of PR created a void, but where there is a demand there will always be supply. Third party companies such as Moz and Majestic have stepped in with their own, in-house metrics such DA/PA and TF/CF. Being not associated with Google or any other search engines these metrics are no doubt less accurate than the original PR. Most importantly, however, while accessing the quality of the link, just like PR, they don’t take into account the relevance of the donor, receiver and the anchor text.

    It must be said at this point, that Majestic must have understood the flaws of their metrics and have been making some attempts to deal with this issue. As a result, they recently introduced a “topical trust flow” which is available to their paid subscribers. This metric attempts to analyse the semantic core of the domain and the donor and assigns a category such as Arts, News, Science, Shopping etc. This is clearly a step in the right direction.


    What does this all mean?

    Can DA/PA and TF/CF still be used while choosing potential link building targets? Of course they can but only as initial filters to find quality sites. To maximise the effect of the link the semantic core of the potential targets needs to be analysed to determine the key word/phrase relevancy of the donor to the anchor text and the receiver. Only then such link can be considered as truly relevant and useful for the user, which in turn will be greatly rewarded by the search engine in terms of ranking.

    It’s important to stress that quality and weight of a link as viewed by Google, Bing, Yandex and other search engines is the factor of usefulness and relevancy for the person following that link. It is this factor that is measured within SE algorithm and gets assigned a numerical value. And it is this factor that directly relates to the position of the site within SERPs for your targeted key phrase.

    Most importantly, a quality link will deliver improvements in SERPs not only for the key phrase in the anchor text but also for the entire group of logically related key phrases. The key phrases within the group are those phrases that statistically are used most often on the web together or in combination with the phrase in the anchor.



    Key term groups in search engines

    Different search engines have different amounts of phrases in associated groups. For example, in Yandex this number is almost 3 times higher than in Google Russia for commercial intent phrases. This is due to lack of understanding of Russian market and inability to create logical connections with all relevant phrases partially because of lack of statistical data. In Google Russia a number of phrases within a group for medium search volume phrases is about 6-7.

    As English is a native language for Google in USA/UK a number of related phrased in a group is substantially greater. This number slowly increases for the higher search volume key terms. This is why an improvement in ranking for high search volume key terms will also bring about improvement for a lot of medium volume search terms. Therefore, high quality links should have high search volume key phrases as anchors because they will have accumulative effect for an entire group of 15-20 phrases. But many experienced SEOs already know that and have been using this knowledge for some time.


    Part 2 will follow shortly:
     
    • Thanks Thanks x 18
  2. seoscientist

    seoscientist Registered Member

    Joined:
    Jul 2, 2016
    Messages:
    86
    Likes Received:
    16
    Gender:
    Male
    Home Page:
    Thumbs up...very good article..
     
  3. PatCook

    PatCook Newbie

    Joined:
    Feb 21, 2015
    Messages:
    17
    Likes Received:
    0
    yes, pr has been dead for a long time, im not sure why people still use it.

    tf is not perfect but can be used as an indicator imo.

    thanks for the read
     
  4. lancis

    lancis Elite Member

    Joined:
    Jul 31, 2010
    Messages:
    1,682
    Likes Received:
    2,426
    Occupation:
    Entrepreneur
    Location:
    Milky Way
    Home Page:
    @tar729 nice article! Havent read anything alike in a long time.

    Although it would be nice to see some formulas here and there. :)
     
  5. RobbinHoodAffiliate

    RobbinHoodAffiliate Newbie

    Joined:
    Jul 6, 2016
    Messages:
    7
    Likes Received:
    1
    Gender:
    Male
    Just joined here and already great to see great advice like this knocking around...
     
  6. seoscientist

    seoscientist Registered Member

    Joined:
    Jul 2, 2016
    Messages:
    86
    Likes Received:
    16
    Gender:
    Male
    Home Page:
    Waiting for Part2 :)
     
  7. T2tkid

    T2tkid Elite Member

    Joined:
    Oct 13, 2010
    Messages:
    3,800
    Likes Received:
    1,731
    Thank you for this write up. Been awhile since I read about something like this.
     
  8. TheUnborn

    TheUnborn Elite Member

    Joined:
    Feb 21, 2013
    Messages:
    3,183
    Likes Received:
    1,722
    Occupation:
    SEO Consultant
    Home Page:
    Excellent, tks for sharing, hope you share some of the methods on how Russian seos are ruling some high competitive niches.
     
  9. tar729

    tar729 Registered Member

    Joined:
    Jun 27, 2016
    Messages:
    90
    Likes Received:
    108
    Why Google discontinued PageRank and why DA/PA or TF/CF is not very accurate (Part 2)



    So let’s go back to the weight on a link and how it is calculated, that very same numerical value of usefulness that determines the power of a link that effects the positions your page in SERPS. All search engines use the same methods to calculate the value of a link in the same way but the quality of data gathering and analysis can differ.

    With time the precision and quality of data analysis improves but the factors being analysed remain the same and, most likely, will always remain the same. This is because this factors and metrics of usefulness are fundamental and are based on fundamentals of human interaction in any society.

    Let’s look at them in more detail:

    A factor usefulness of a link for a user can be determined by the links click through rate. Obviously, if an anchor is relevant to the donor and receiver pages the click through rate will be greater. The position of a link within the donor page also makes a difference. So when users visit a donor page and click on a link they are hoping to find potentially useful information. Therefore, CTR is a great indicator of how relevant and useful the link is, or at least perceived to be.

    On the other hand, CTR can only indicate perception but not the actual relevance. If, for example, you visit a site selling hardware and tools and click on a link with anchor “electric drill” and land on the page about dog food, most likely you will be disappointed and confused. Therefore, the second metric of usefulness is the relevancy of the anchor text to the content of the receiver page. However, if the content of the donor and receive pages are very similar the user is unlikely to receive much benefit. Even if the receiver page is more relevant to the anchor then a donor page the user will not receive the information they seek and quality of their search will suffer.

    With this in mind, search engine algorithms attempt to not only measure the relevancy of the anchor to donor and receiver pages but also to determine the differences in the information presented there. It is at this point that the ability of a search engine to accurately measure the usefulness of a link directly depends on the ability of the algorithm to extract meaning from both donor and receiver documents

    If some of you still think that Google is unable to understand the meaning of a page, I will not try to prove you otherwise here. I will just say that this is not the case. I might at some point cover this topic in more detail and talk about semantic graphs and automatic translation algorithms (that by the way use the very same technology rather than simply replace words with translations individually).

    Another important metric while measuring the quality of a link is traffic on the donor page. No matter how relevant the link is it is not very useful if the donor page has no visitors. However, it must be said that SE algorithms do not measure actual traffic but estimate it based on the position of a page within SERPs. Any data Google might have related to actual traffic of a page are not used as a ranking factor. This information forms the main body of statistical data which is used by the search engines to improve automated learning and correct any inaccuracies in traffic estimates.


    Of course, this process of weighing up and assigning a numerical value to each link is a very complex process and currently the margin for error is still quite high. However, over time it will improve.

    Needless to say, DA/PA and TF/CF are not related to this metric in any way, shape or form which makes those indicators incomplete and somewhat outdated.

    In Russia SEO specialists already have tools that allow them to measure links’ weight using the logic and methods described above. No doubt this technology will make its’ way here too but it might be few years before it does.

    I hope this helps.
     
    • Thanks Thanks x 3
  10. Logan.J

    Logan.J Registered Member

    Joined:
    Jun 10, 2016
    Messages:
    91
    Likes Received:
    14
    Gender:
    Male
    Location:
    Behind
    Top notch post.Keep sharing like this :)
     
  11. VladReverse

    VladReverse Newbie

    Joined:
    Jul 5, 2016
    Messages:
    2
    Likes Received:
    1
    Gender:
    Male
    Occupation:
    CEO & Developer
    Location:
    Romania
    Thank you for the share! This was a good read.
     
  12. Zerzar

    Zerzar Junior Member

    Joined:
    May 27, 2014
    Messages:
    127
    Likes Received:
    9
    Gender:
    Male
    Thanks for this share.
     
  13. turdface

    turdface Supreme Member

    Joined:
    Sep 3, 2010
    Messages:
    1,414
    Likes Received:
    184
    That was a decent write up that considers more than what is usually discussed.
     
  14. LatteGrande

    LatteGrande Jr. VIP Jr. VIP Premium Member

    Joined:
    Jan 19, 2011
    Messages:
    2,197
    Likes Received:
    613
    Location:
    404 Not Found
    That's pretty refreshing stuff. Keep it up OP.
     
  15. sabrinar2

    sabrinar2 Registered Member

    Joined:
    May 7, 2015
    Messages:
    52
    Likes Received:
    1
    Occupation:
    Student
    Location:
    USA
    Thanks for the share OP! Please keep it updated.
     
  16. WebTG

    WebTG Jr. VIP Jr. VIP

    Joined:
    Mar 18, 2015
    Messages:
    578
    Likes Received:
    149
    Brilliant, this is what makes BHW forum best
     
  17. blogtaufiq

    blogtaufiq Junior Member

    Joined:
    Mar 22, 2011
    Messages:
    147
    Likes Received:
    10
    this is a good article. better than your previous article. now i understand what you mean about donor, receiving and semantic. thanks
     
  18. Gallag97

    Gallag97 Junior Member

    Joined:
    Jun 16, 2016
    Messages:
    182
    Likes Received:
    21
    Gender:
    Male
    a very unique article, i like it
     
  19. tensile

    tensile Registered Member

    Joined:
    Jul 16, 2015
    Messages:
    90
    Likes Received:
    53
    Location:
    Switzerland
    tar, very interesting stuff, but what would be great would be your opinion about what online marketers should do with that info. If I understand correctly, it seems like we should aim for links based on the highest level search terms and not using long tail phrases (since the higher level links feed through to lower level search terms anyway).

    But on a practical level what else would you suggest to maximise our SEO efforts?
     
  20. tar729

    tar729 Registered Member

    Joined:
    Jun 27, 2016
    Messages:
    90
    Likes Received:
    108
    to measure the weight of links
    I'm an owner and developer of one of those services in Russia http://levsha-instruments.ru/, but the measurement algorithms don't work on English links yet. I need an investor to develop such service for your market. Investor gets 10% share. For more info contact me via skype: ev-gen.t
     
    Last edited: Jul 9, 2016