Today we are announcing the first major update to the LinkRisk algorithm of 2017. As always with our updates we like to name them after someone in the industry and today’s update is no different.
This update is called LinkRisk Dance ‘Badams’ after prominent SEO, our good friend, and KTOC Alumni Barry Adams. Barry runs the multi award winning agency Polemic Digital based in Northern Ireland (https://www.polemicdigital.com/). Barry recently scooped 2 awards at the UK Search Awards (https://www.polemicdigital.com/uk-search-awards-wins-polemic-digital/). A hugely popular choice on the night and something the whole team at Polemic can be massively proud of, we know very well that winning at the UKSA is not an easy thing to do.
You may have noticed I referred to Barry as a ‘KTOC Alumni’ too. KTOC is the rather informal and fun annual competition day held in Las Vegas each year (KTOC stands for the Kerboo Tournament of Champions). I mention it mainly so I have an excuse to include this picture of Barry from that event, I think its pretty representative of how it normally goes!
Now on with the update information.
Update Badams covers a couple of important changes and puts in place some more infrastructure in the backend to allow us to make faster and more responsive updates to the scoring of our LinkRisk metric.
New domain sensitivity and repeated pattern weighting.
During review of the data we see in the profiles run by clients and in our own managed service work certain patterns emerge. The system is trained to take these highlighted patterns and apply certain rules which influence the score. An example of that is the way the system handles some of the new domain extensions like .xyz
It is pretty obvious to anyone who has done a link audit recently that the vast vast (if not all) of the links from .xyz domains are spammy and problematic. The system has now had that pattern highlighted and is therefore more sensitive to them and applies risk accordingly.
Classification methods to help aggregate similar risk profiles.
Another way we can deal with matching datasets within our scoring is to apply certain classifications to sites and then compare the classification a site has been given with other sites that also have that marker. This allows us to make the system more consistent in how similar sites are scored.
Infrastructure
Finally this update continues the drive at Kerboo to add to our already complex back end infrastructure. Our updates to the LinkRisk scoring algorithm are named but that doesnt mean that small adjustments dont happen on a daily basis. This update allows for even more granular updates in real time to the scoring system making the system more accurate and responsive to changes as they occur, building on the work completed during LinkRisk dance ‘Jesse’.
How will this update impact my existing LinkRisk scores
This is a fairly large update of the scoring system but the updated scores will only be calculated on a rescan of your profile.
We would suggest that every domain you look after on a regular basis is either updated automatically on a monthly basis by the systems automated rescan option or at least quarterly using a manual rescan.
It is recommended that anyone in the following markets or with the following issues also considers a manual rescan of their data to benefit from this update: –
Anyone with a significant issue with repeated spam from new domain extentions or from sites with a repeatable pattern.
Anyone in Fashion, Home & Lifestyle or retail / consumer goods.
The post Algorithm Update – LinkRisk Dance ‘Badams’ appeared first on Kerboo.