Google published, in August 2018, the latest version of its search quality evaluation guide . A 160-page guide that I have condensed here into an easy-to-understand infographic .
These guidelines are representative of what search engine users want to see as quality results .
Webmasters and agencies can ensure that their web pages and content receive the highest possible quality score to optimize their SEO .
The infographic on content quality according to Google guidelines
A page can get a high rating without reputation, but the opposite is impossible (a high score with a bad reputation).
Content from a young site (less than 2 years old) with no reputation cannot be considered low quality, however, older sites should have reviews.
Measure the quality of a page
Use this table to identify areas to improve in your content or pages to get a better rating.
Very low Low Half High Very high
Content Objective (The Promise) Failed Failed Accomplished Accomplished Accomplished
Reputation Negative Negative Positive Positive Very positive
Main content (MC) size Very short Short Half Long Very long
Quality of the main content (EAT Principle) Low quality Mediocre quality Satisfactory quality Good quality Excellent quality
Secondary content (SC) Asia Pacific Lead Telemarketing Distractor (abusive advertising) Distractor (abusive advertising) Absent Useful Very useful
Web design Poor Poor Poor Functional Functional
Information about the website Absent Absent Absent Present Present
It’s easy to imagine that Quality raters have a fairly similar matrix for evaluating content .
Does Google AI use data from Quality Raters?
We know that Google uses machine learning at many levels in the search engine. This includes Google RankBrain , which is part of the algorithm and helps Google rank results for the 20% of daily queries that Google has never seen before.
Every time Google publishes new recommendations for quality assessment. One might rightly wonder whether the search engine integrates the results of Quality Raters into its Machine Learning algorithm.
My opinion on the subject is that the sample Hong Kong Lead size , that of the evaluators, is too small to constitute a solid data set for the creation of the dataset.
The data sets required for the use of machine learning tend to be quite large, and even if there is a significant number of quality evaluators (more than 10,000 worldwide), it is far from the number to constitute a robust data set to fulfill the search engine’s mission.
