MozRank Per Anchor

Links with targeted anchor text are generally regarded to be the most important factor in search rankings according to search engine optimization experts. However, this factor has remained particularly difficult to nail down as we have had little to no access to the data necessary to correlate anchor text links to actual rankings.

Historically, the standard measurement we have had available for determining the extent to which a page was externally optimized was a simple ratio where the number of links pointing to a page with the correct anchor text was divided by the total number of links pointing to that page and then multiplied against the PageRank of that target page. This would allow us to determine what percentage of PageRank could be attributed to the correct anchor text. Below is a pie chart of exactly that type of measurement. Each slice represents a different piece of anchor text. As you can see, approximately 42% of the anchors pointing to this page are using the correct, targeted anchor text.

The Link Profile by Anchor Text without any Weighting Based on Link Quality

The best metric used to be looking at the percentages of anchor text usage, but this technique was problematic at best as it could not weight links based on quality.

This data is problematic, though, as it only looks at Anchor Text Quantity and not the Quality of Links with said Anchor Text. More astute SEO’s would work out slightly more complex formulas to determine this data. For example, we could look not only at the anchor text of each link, but the PageRank of the page on which that link originates. We can then add up all the PageRanks for each anchor text and generate a new graph. Unfortunately, this data is problematic as well. How do we know the amount of PageRank passed by each link on the page? The link may sit on a PR6 page, but that same page may have over 1000 links on it. How do you transform the PageRank into a non logarithmic scale so that the numbers can be accurately compared?

This is where the new Site Intelligence API from SEOMoz comes in handy. The API gives you access to tons of data but, perhaps most importantly, gives you access to the Raw MozRank passed by each link. With this data, we can better understand what I would consider to be a holy grail metric – MozRank Per Anchor.

As you can see in the follow up graph below, we have charted the same set of anchors used on a site. However, this time we have determined percentages based on the aggregate of MozRank passed in each link for each keyword. With this data combined, the numbers look quite different. Only 30% of the MozRank is passed by links with the appropriate anchor text. In fact, one of the important keywords accounts for 8% of total anchors, but only 1% of MozRank per Anchor. The disparities are startling and help illuminate why good data is essential.

The Link Profile by Anchor Text when Weighted by Link Quality Looks Very Different.

Now we can consider anchors based on the quality of the link rather than simply the quantity.

This is extremely important for clients who are trying to spend their link budget efficiently. The client can now identify that one of the keywords they are targeting is significantly lacking, while another represents a greater share of the MozRank than it ought to. The client should alter his/her link building campaign to target the poorly represented keyword by pulling back on the over-represented keyword. Moreover, the client can compare his/her MozRank per Anchor vs competitors to determine which gaps are surmountable.


  1. The number of links with the correct anchor text is meaningless if those links do not pass PageRank
  2. Using the Site Intelligence API from SEOMoz, we can pinpoint and aggregate the MozRank passed by each link with particular Anchor Text
  3. Knowing the MozRank per Anchor lets us know how to craft a campaigns more efficiently.
No tags for this post.


  1. Michael Martinez
    Nov 4, 2009

    But you have no way of knowing which links are passing value within any particular search engine’s index.

    Linkscape is no more helpful than Yahoo! in assessing the value and quality of links as another search service (like Ask, Bing, or Google) sees them.

    Having said that, I think your methodology at least helps to bring less experienced SEOs up to the intermediate level — where they at least have a tool to make some sort of guestimate.

    It’s better to take a carefully considered shot than a blind shot even if you don’t have all the data.

    Author Note: As always, thanks for the good response Michael. You are right – Linkscape does it’s best to model against PageRank and even I have noted some of the earlier disparities. However, you are also right in that this gets us one step closer. I do disagree with you that Linkscape is no more helpful than Yahoo! in assessing the quality of links in another search service. Yahoo! gives you no more data than a mere list of backlinks – a list which is presumably ordered by importance but does not show any relative difference. I do not pretend to believe that MozRank/Linkscape can serve as a perfect predictor, however it is the best thing we’ve got – and the closest thing to objectivity in an industry rife with disagreement. Thanks again for commenting, as always.
  2. Michael Martinez
    Nov 5, 2009

    The problem with using LinkScape or any other tool to try to guess which links in Google pass value is that the third-party tools have no insight into which filters or penalties (if any) are affecting links.

    LinkScape might make it easier to say “this page should probably pass value” but it can never justify saying “this page will definitely pass value”.

    As I indicated above, it’s preferable to have a “probable” guess to being stuck with an “i dunno” guess but in the end it’s still guess work.

Submit a Comment

Your email address will not be published. Required fields are marked *