Link Score Introduction
In version 10 of the SEO Spider, we introduced our own Link Score algorithm to help users identify the relative importance of URLs based on the internal linking architecture of a site.
It can be considered as an internal PageRank calculation on a user-friendly 0-100 point logarithmic scale. The higher the the value, the stronger the page. Link Score can be viewed in the respective ‘Link Score’ column in the Internal tab, as shown below. The Link Score algorithm can take some time to run and relies on a stable set of data to work with. For this reason Link Score is performed at the end of a crawl as part of Crawl Analysis, which can be performed by clicking ‘Crawl > Start’ from the top level menu.
Identifying Eligible URLs
The first stage in the SEO Spider Link Score algorithm is to identify the subset of URLs from the crawl that are eligible for Link Score. This is all URLs that meet the following criteria:
- Not a redirect.
- Not canonicalised.
- Linked to either by an AHREF, or are the target of a redirect or canonical.
Initial Link Score Assignment
With the eligible URLs now in place, each one is assigned an initial Link Score of 1/n where n is the number of eligible URLs. The is just a start point before the SEO Spider allows Link Score to ‘flow’ between URLs.
Passing Link Score Between URLs
Next, Link Score is passed between the URLs based on link relationships. We will discuss what links are considered later, but for now we’ll focus on the main part of the algorithm.
Each eligible URL is considered in turn and has its Link Score updated. Every inbound link passes Link Score. It passes the Link Score of the originated URL, divided by the number of outbound links. The total Link Score from each inbound URL is then added up and used to set the URLs Link Score as follows:
Link Score = ((1 – D) / n) + (D * total link score from in links)
Where n is the number of eligible URLs and D is the Damping Coefficient. The Damping Coefficient controls how much Link Score flows between URLs. The SEO Spider uses the commonly accepted value of 0.85 for this calculation.
The act of updating each URLs Link Score once is an ‘iteration’. The 1st iteration uses the Link Score initially assigned. Subsequent iterations then use the Link Score from the pervious iteration. After a number of iterations the Link Score values settle down. The SEO Spider performs 10 iterations in total.
Not All Links Are Equal
Not all links are considered for Link Score. Only the following link types are considered:
- Canonical (HTTP/HTML).
- Redirect (HTTP/Java Script/Meta Refresh/HSTS).
Links must also be:
- Non self referencing (A URL can’t gain Link Score by linking to itself).
- Unique (Multiple links from URL A to URL B are ignored).
- Followable (not rel=’nofollow’).
The Link Score algorithm has some specific rules around certain links types. A ‘nofollow’ link causes Link Score to ‘evaporate’, in a similar way as it’s been suggested occurs to PageRank. Nofollow links do not pass any Link Score, but are included in the total number of outlinks calculation, hence the ‘evaporation’ of the Link Score.
Redirects and Canonical links have the same effect as each other in the algorithm. Link Score bypasses the redirected or canonicalised URL to its target URL. For example, if URL A redirects to URLS B, all in links to URL A are considered as being inlinks to URL B.
The Link Score values themselves aren’t very useful, as the larger the site, the smaller the values and they can be difficult to interpret easily. Therefore, to make the results easier to work with, the SEO Spider maps these values to numbers in the range of 1 to 100. The URL with the highest Link Score is given value of 100, the URL with the lowest Link Score is assigned a value of 1, with values in between scaled to this range.