Chris Long’s Post

Huge SEO news: A new Google exploit was discovered that revealed 2,000+ classifiers including click probability, site_quality, consensus + more: At SearchNorwich, Mark Williams-Cook gave a session called " Improving your SEO with conceptual models". This session discusses a Google end-point that Mark discovered that let to him analyzing 2,300+ Google properties and potential algorithm factors. This is extremely similar to the Google API Leaks that were discovered earlier this year. Mark's session contains a lot of great insights from the Google exploit. This includes findings such as: 1. Consensus: Google may look to understand the consensus of both a page and an entire query type. There are variables such as "consensus_num_passages_agree" & "consensus_num_passages_contradict" that determine whether or not your content agrees or disagrees with previous statements you've made. 2. Newsworthy Queries: Google also has a classifier that determines if a query is related "news". Mark found there was a variable called "newsiness_score". 3. Medical Queries: Mark also found that there is a classifier that determines whether or not a query is related to a medical term. 4. Intent Classifiers: Google classifies all queries into one of eight types: SHORT_FACT, REASON, DEFINITION, OTHER, COMPARISON, INSTRUCTION, CONSEQUENCE, BOOL. The vast majority of queries he analyzed are either SHORT_FACT or BOOL (true/false). 5. Expected CTR: Google also builds an expected CTR for a given query in a click_age_probability classifier. There is a tool in the Google Ads Planner that can give you an estimated CTR of a given query. 6. Site Quality: Mark found that Google has a site_quality score associated with every site. This is actually measured at a SUBDOMAIN level. Most sites fall between a 0.6-0.7 measurement. 7. Measurement: Mark went and found the "Site Quality Score" patent that Google filed. This is measured using factors such as how often people search for your site, how often people select your site, how often people use your site in anchor text Similar to the Google API Leaks released back in May, this exploit gives us additional data about Google's algorithm. Mark and his team were given $13,337 reward from Google since the vulnerability they discovered was high priority.