Jonathan Leger – SEO And Internet Marketing Blog Internet Marketing Blog


What 456 billion links failed to show has released the conclusions they've reached from analyzing the hundreds of billions of links and web pages they have indexed in their massive database, Linkscape.

There's not much new there, which is no surprise, though it's nice to have some hard evidence support the conclusions I've drawn from analyzing Google's search results for the last few years.

I recommend you read the whole post, but here's the gist:

1. Get links from unique domains to your site.
2. Have the keywords in the domain name.
3. Have the keywords in the title.
4. Don't overdo it with the keywords.

Some of the common myths are busted in there too:

1. PageRank is the holy grail (the data shows it isn't).
2. On-page factors matter a lot (except for the page title, no they don't).
3. Subdomains with the keywords help (nope).

However, SEOmoz is failing to discern a major component of why Google's search results look the way they do. They say in their blog post regarding their results that they don't have the whole picture, because it's clear from their data that although getting links to the ranking page shows a real correlation with the page's ranking, it's not the whole story. The page with the most links doesn't always win.

I am floored that they don't know the reason why, despite their obvious technical capability of building such a massive, sophisticated database.

What they are missing is this: they are not taking into account the ranking domain's authority (the number of links to the WHOLE site, not just the ranking page). They were only analyzing the number of links to the ranking page itself.

I bet if they took into account the total number of links coming into the ranking site they would quickly see why the SERPs aren't based solely on the number of links to the ranking page--some of those results have lots of authority despite having few links to the ranking page.

Wikipedia pages, for example, rank primarily because the Wikipedia site has hundreds of millions of links (198 million according to Yahoo! Site Explorer as of right now), which in Google's eyes makes the inner pages trustworthy to rank even though they have few links themselves. This happens with lots of other sites, too (EzineArticles, Amazon, etc.).

I can't believe SEOmoz is even looking at H tags (H1/H2/H3 etc.). H tags have had virtually zero impact for many years.

Let me make it real simple for you. To rank for anything in Google, all you have to do is:

1. Register an exact-match .com/.net/.org domain name
2. Include your keywords in the title tag.
3. Get lots of links from unique domains


4. Don't be surprised if Wikipedia still outranks you (unless you can come up with 200 million links yourself).

It's not that complicated. That's obvious to me from years of analyzing Google's search results. And I didn't need a database of hundreds of billions of links to only come to a partially correct conclusion.

Please post your thoughts and questions in a comment below.

Related Internet Marketing Q&A

Comments (41) Trackbacks (0)
  1. Its been about 6 months since you wrote this post, do you still feel the same about on-page ranking factors?

    It seems to me that even though h1 tags for instance *may* provide some increase, even that little tiny bit could help in the grand scheme of things.


  2. I don’t think anyone thinks PR is the holy grail any more, this has been disproved time and time again on various forums

  3. Hi Jon.

    firstly, I wanted to say I have huge respect for you, and I consider reading your “Search Engine Myths” as a turning point in my SEO career back in 06/07, whenever that was.. .

    that said, I now have to disagree with you about numbers of links winning out, its clearly about quality of links now, and internal as well as external links.

    if it was pure numbers, to either the page or the domain, then there’s no way we’d ever be outranking wiki is there?

    and we do, all over the place .. ;)

  4. Just subscribed to the blog last month, keep the great updates coming.

  5. That insight about Wiki makes me think, I should start Branding myself…get 1/2 million links thru my different topics or subdomains and then I can top Wiki….
    …well any other ideas?

    Do you find Video helps jump you ahead?

  6. You have alot of good points in this article that I agree with. When it comes to links you can go a far way ranking with your website and choosen keywords. Sweet Article!

  7. For the past several weeks I have been researching non-stop about web design, SEO, and linkbuilding. I have found a lot of useless information, some helpful information, and then your website. I honestly can’t remember the last time I read so many posts from a single blog. So, thank you.

    I’m looking forward to using your blog search tool, and I just made my first KeywordLuv comment!

    Thanks again, :-)

  8. Aside from that, I think domain names can give you a boost, I just don’t feel it’s as strong of as boost as you feel it is. I also think that boost can be quickly overshadowed by other SEO factors. I think the same of anchor text, I think there is a limit to how much anchor text affects rankings. Obviously, I’m trying to figure all of this stuff out and don’t have all of the answers but when I dig into the big sites a lot of SEO myths appear to be thrown out the window and the methods that work for low competition keywords aren’t the same as what is working for the big sites.

  9. Thanks for clarifying the key points, Jonathan.

  10. I cannot believe there are that many links. Every one of those links was created by someone’s thought. It might have been that then it was created by automation but wow that is a lot of links.

    How many have you created Jonathan?

  11. Thanks for the summary Jon.

    I’ve picked up on the importance of a exact match domain from you but I’m a little surprised about your almost complete dismissal of on page SEO

    BTW does your analysis include anchor text for the links?



  12. You had a great insight here but it looks like Rand Fishkin dispelled your misconception about the factors they were looking at. Still, I think you’re mostly right about the main factors that go into ranking well. It’s as simple as you say it is, as long as you’re talking about a relatively uncompetitive search phrase. If you’re trying to rank for a phrase saturated by SEO companies with large budgets, it takes more than an exact match domain and a title tag to rank on the first page.

  13. Wow!! it s really a great news that has released 456 billion links…….It is really a great sharing,,,,,

  14. >>1. PageRank is the holy grail (the data shows it isn’t).

    Heard that Google is doing away with the PR thing. But I know many people are obsessed with it.

    > >2. On-page factors matter a lot (except for the page title, no they don’t).
    I agree with this. Hardly do any onpage SEO stuffs except the title.

    >> 3. Subdomains with the keywords help (nope).
    I always thought this will help in the ranking. Guess it is good news cos one less thing to try.

  15. Hi Jonathon

    Great info on your site but I have a question.
    If ranking on Google is as easy as you say on the latest post about SEOmoz – why would I need your program to get ranked on Google.
    I’m not being a smart ass, just asking a real question.
    I like the way you address these issues and I’m looking for some help with two new websites..
    Regards, Larry Allen

  16. I definitely agree that they need to look at the number of links the entire domain has. Domain link popularity can often trump the link popularity of an individual page. If you’ve ever gotten to the top of the search rankings, you know how easy it is to then get rankings for other pages on your website.

  17. Jonathan,
    I remember listening to your very informative SEO interview with Keith Baxter few years ago and you have said most of the things that people are still discovering or don’t to believe it because the getting the right links are really hard to control and /or time consuming.

  18. Thank you for the post, I’ve found a lot of helpful information here!

  19. Great post. thanks for this I enjoy it very much not only this post but some comments are also very nice. thanks to all of them.

  20. Hi, jonathan, You have said one thing in this post that the Keyword should have in headline. I would like to ask you that what is the benefit of that.

  21. Wikipedia pages, for example, rank primarily because the Wikipedia site has hundreds of millions of links (198 million according to Yahoo! Site Explorer as of right now), which in Google’s eyes makes the inner pages trustworthy to rank even though they have few links themselves.

    Isn’t the reason that wikipedia does so well for inner pages, that all its internal links are ‘follow’ (while external are all nofollow), so it’s continually giving it’s own pages ‘linkjuice’ boosts

  22. Your site is very nice and the work at SEOMoz is pretty interesting from an academic standpoint, but as you correctly point out, the 80/20 rule applies well here. You can get 80 percent of the benefit (or more) by understanding the simple “20 percent” that you outline above.thanks..

  23. Pages and links, pages and links,

    Simple analysis is sometimes better than analysis done by a huge volunteer organization. Would you expect anything enlightening coming from DMOZ?


  24. You said right my friend if we have to reached at better conclusion so have to read whole post that is way i have bookmarked this site for reading it briefly.

  25. If the exact domain name, already register, so how the solutions? we add a name front or at back or register .biz .info and etc?

    Thanks for explanation

  26. I guess they didn’t have enough data. :)

    There is one more way that can rank real well. I’ve seen it done and it works quite well. The G00gle bots don’t do too well with tables, and using tables inside of tables, to make it appear you have far more pages than you really do.

    That’s it though, not giving any more clues.

  27. Great post Jon.

    How competitive of a keyword are we talking about here? Depending on the competition, links of any kind may be enough to get you to the first page on search results..

  28. Great post, although I do not agree with lots of links from unique domains. One of my clients recently have been awakened to the reality by a web site less than a year old that just had purchased throughout links (i.e. links from every page on a site) from couple of web sites and immediately skyrocketed to the middle of the first page. Looks like the sheer volume of links can make up for value of domains that link to you.

  29. Great summery on the report Jonathan!

    There is still so much of the old SEO information being taught and touted as the ultimate truth. Your four points might potentially put a lot of SEO experts out of business. You should be careful – lol.

    Hey, thanks for such a great deal on Answer Analyst; its like having Christmas in November. I’ve been tinkering with it and so far I’m loving it. Nice piece of software. You are the man!



  30. hi jon!

    somebody that are doing xfactors adsens building says .info works great…any testing on this yourself?

    how about your new domains for this MFA sites, how fast do indexing take? are you only doing the backlinks from your own service then?
    asking because heard many uses old domains to rank high but then i cant choose exakt keywordphrase domain so…


  31. Low hanging fruit, high no. of searches alongwith a perfect matching
    Keyword Domain with an .org extension is the key to be in the game.

  32. As you indicate, not much new. However, I wasn’t clear on the reference to Wikipedia and EzineArticles inside links not having much value. I try to write several articles a week for Ezine, and put links in resource boxes. I notice most of the pages I see in Ezine do have some type of link. Bottom line: I’ll have to do some research about the inside links of Ezine

  33. I’ve always wondered why you see internal pages on sites like Amazon and Wikipedia ranking so high. They have so many links that they are authorities in many different niches.

  34. Good post. It was good to read Rand’s responses too.

  35. As usual, great post Jonathan.
    Domain authority certainly weigh more than anything else as far as the value of the inbound link is concerned. Anchor text and the relationship between the two sites linked is also important because it shows the relevancy of that link.

  36. Each graph is showing different metrics, but in the ranking model used to find causality (which comprises the latter half of the post), features about links to a domain are absolutely a part of that. In the raw correlation models at the top, we’re only looking at a single metric in isolation, but that’s the intent – to see what the direction correlation is between a single metric and rankings. The much more complex model using hundreds of features (and thousands of derivatives) attempts to show the actual impact of a single factor (causality, not just correlation) and it does use all those metrics I described above.

  37. randfish:

    Thanks for dropping by and leaving a comment. I’m rather confused though — are you, or are you not, taking into account the total number of links into the ranking domain? Because the blog post makes it sound like you are not, like you are only taking into account the number of links to the ranking page.

  38. Thanks for this great post Jon. The work at SEOMoz is pretty interesting from an academic standpoint, but as you correctly point out, the 80/20 rule applies well here. You can get 80 percent of the benefit (or more) by understanding the simple “20 percent” that you outline above.

    I wish they would have spent more energy understanding the impact of having inbound links with the target keywords in the anchor text. But, as you point out, without considering the authority of the ranking domain you probably cannot see the correlation…

    Thanks for another great post.


  39. Hey Jonathan – thrilled you read the post and got value from it. Just FYI – we most certainly do use lots of metrics about a domain’s authority and value. Specifically, we calculate and show the following via LInkscape:

    - Domain mozRank
    - Domain mozTrust
    - # of linking root domains
    - # of total links
    - distribution of anchor text to a domain
    - distribution of (the many different) metrics about links to a domain
    - distribution of (the many different) metrics about a domain with a link to another domain

    Our ranking models use these data points and many others (and derivatives of them) to build the causality estimates you see at the end of the post. Don’t worry – we wouldn’t miss something obvious like that :-)

    Thanks for the post!

  40. Great post as usual Jon. Sometimes the simple things like looking at a site on a macro level rather than micro page at a time is something that is easy to overlook.

    I know my niche sites when I have keyword domain, title and different links works like a charm and usually gets me indexed and clicks within 24 hours.

  41. Thanks for the interesting post, was a great read. And you’ve made it really simple (at least conceptually) to focus on the 80/20.

Trackbacks are disabled.