If it's spam, why is it even being rated?
Posted: Tue Feb 18, 2025 6:44 am
For Google, evaluation is a multi-step process. Time, data access restrictions, and cost prevent the implementation of more complex systems. Of course, the assessment of documents never stops. That is why the pattern repeats itself. Content first goes through an initial “sniff test” before being identified later.
We can look at some evidence for this claim. A spain mobile database while ago we looked at Google’s “Page Quality” patent and how it uses user interaction data to generate a ranking score.
In a situation where a website is brand new, users have not interacted with the content on the SERP. Therefore, Google does not have access to the quality of the content. However, another clever patent covers this situation, namely Predicting Site Quality . To simplify: for new websites, the quality score is predicted by first obtaining the relative frequency for each of the different phrases found on the new website. These measures are then mapped to a previously generated phrase model built from the quality scores created from previously ranked pages.
If Google were to continue using this tactic, it would mean that many new websites would be ranked based on a “first guess” with quality metrics built into the algorithm. Later, the ranking is refined based on user interaction data. In practice, we have noticed that Google often increases the ranking of websites for what appears to be a “testing period.”
We can look at some evidence for this claim. A spain mobile database while ago we looked at Google’s “Page Quality” patent and how it uses user interaction data to generate a ranking score.
In a situation where a website is brand new, users have not interacted with the content on the SERP. Therefore, Google does not have access to the quality of the content. However, another clever patent covers this situation, namely Predicting Site Quality . To simplify: for new websites, the quality score is predicted by first obtaining the relative frequency for each of the different phrases found on the new website. These measures are then mapped to a previously generated phrase model built from the quality scores created from previously ranked pages.
If Google were to continue using this tactic, it would mean that many new websites would be ranked based on a “first guess” with quality metrics built into the algorithm. Later, the ranking is refined based on user interaction data. In practice, we have noticed that Google often increases the ranking of websites for what appears to be a “testing period.”