Here are the top 20 takeaways from the Google API leak
How to Utilize the Google Search API Leak for Your Own Optimization
Last Friday, information about the Google Search API was accidentally released. This provided some insights. Rand Fishkin and Michael King quickly turned this information into detailed articles.
The SEO community is pleased that Google inadvertently made thousands of pages of internal documentation on search engine technology public. These documents could provide valuable insights for the SEO industry. The information contradicts some of Google's official statements about the fundamentals of their search results. The leaked documents pertain to Google's "Content Warehouse API." They list over 14,000 attributes used for weighting search results. However, the specific weighting of these attributes is not disclosed. Google has not yet commented on this leak.
While both Rand and Michael report in detail, one question remains: What are the key takeaways from these leaks and how should they be evaluated?
- Modular System: Google uses 2,596 modules and 14,014 attributes to evaluate websites.
- Domain Authority: "siteAuthority" is indeed used, contrary to Google's previous statements.
- Click Signals: User behavior and clicks influence rankings, contrary to Google's earlier claims.
- Sandbox Effect: New websites can be isolated to prevent spam.
- Chrome Data: Data from the Chrome browser is included in the evaluation.
- Link Evaluation: The quality, relevance, and context of links are crucial; bad links can have negative effects.
- NavBoost: A system for utilizing click and behavior data.
- Whitelists: There are whitelists for specific areas like COVID-19, elections, and travel websites.
- Quality Raters: Feedback from human quality raters is used.
- Link Categories: Click data determines the quality of link categories.
- E-A-T (Expertise, Authoritativeness, Trustworthiness): These principles are central to assessing page quality.
- Content Freshness: Current content is preferred.
- Page Experience: Signals like loading time and mobile-friendliness are important.
- RankBrain: Machine learning is used to better understand search queries.
- BERT Algorithm: Used to capture the meaning of words in search queries.
- Subdomains vs. Subfolders: Google treats subdomains and subfolders largely the same, emphasizing specific implementation.
- CTR and Dwell Time: Although Google publicly states these signals are not used, the leaks suggest they may play a role.
- Mobile-First Indexing: Google primarily evaluates the mobile version of a website.
- AI and Machine Learning: These technologies play an increasing role in evaluating websites.
- Personalized Search: Search results can be personalized based on user behavior.
What Do the Documents Say About Disavow and Link Audits?
The documents do not specifically address disavow and link audits. Disavow data may be stored elsewhere, not in this API. Since quality raters' data is directly accessible, it suggests that disavow data is decoupled from core ranking systems. It is assumed that disavow was used as a source to train Google's spam classifiers. The fact that the data is not "online" supports this assumption. Google has refined link analysis with features like IndyRank and PageRankNS. It is advisable to review your link-building strategies in light of this information.
- Disavow Links: The disavow tool allows webmasters to ask Google to ignore certain backlinks. This is used to neutralize potentially harmful links that could lead to ranking penalties. The tool should be used cautiously and only for clearly identified harmful links (Ahrefs) (Semrush).
- Link Audits: Regular link audits are important to monitor a website's link profile and identify harmful links. Tools like Google Search Console and specialized SEO tools (e.g., Semrush, Ahrefs) help find suspicious links and create a disavow list if necessary (Semrush) (Semrush).
- Toxicity Score: SEO tools often rate links with a "Toxicity Score" to determine the quality of backlinks. A high score indicates harmful links that may need to be disavowed (Semrush) (Semrush).
- Recommendations for Link Removal: Google recommends removing harmful links directly from the linking website before using the disavow tool, although this can often be difficult (Semrush) (Semrush).
- Manual Actions: Disavow should be considered especially if a manual action by Google due to unnatural links has been imposed. Without such an action, the risk of a penalty is lower, but monitoring remains important (Webmasters Stack Exchange).
These insights offer a deep look into the complexity and numerous factors influencing Google's search algorithm.
