Basic Concept

LinkRisk is our attempt at deriving the intent behind the link placement and as such is very dependent on the factors observed when the link is actually live. We take 150+ signals (which come from data partners, crawling the sites, 3rd party APIs and our pool of data gathered from users of the system) and run them through 125+ different rulesets - the output of each serving to nudge the score up or down from its base of LR500 until a final score is arrived at.

Signals

Some of the signals we gather include (not a complete list): -

- Link metrics (Majestic / Ahrefs etc.)
- Status codes of pages
- Language of pages
- Hosting information
- Contact information
- Whois information
- Text from the sites
- URL patterns
- Common footprints
- Where the link is on the page
- Social activity for the site

We also gather information about the people involved with the sites (personas; not an author but someone who has activity on the site, they might own it or they might be a regular commenter etc. etc.).

What we also know

We also know:
- If the site has been disavowed by people in Kerboo
- How people have rated the site during investigate
- Known sites from over 10 years of link building
The way the algorithm uses those signals is to set every link to a base score of 500 (Neutral) and then run 'rules' through the algo to nudge the score up or down.

Example

It's probably easiest to explain that through an example: -

Say you’re a UK based site.
- The system sees you have a link from a .pl domain. This nudges the score up very slightly.
- The system notes that the link is from a forum - this again nudges the score up slightly.
- The system spots that the page linking is in English rather than Polish. This nudges it slightly down again.
- The system spots that the link is in a profile page rather than a post - this is a known footprint for automated forum spam and so the score nudges strongly towards risk.
- The system notes that the domain has actually been disavowed nearly every time its been seen by other users. This then pushes it firmly into the high risk band.

Want to know more?

Get in touch with your questions and we will get back to you promptly with the answers and advice you need.