Johannes Beus
(Author)
23.05.2025
After US courts have already ruled that Google abuses its monopoly position, the question that now follows is which conditions should be imposed in the future to prevent this. In this phase of the procedure, interesting insights for SEO are coming to light as well.
Recently, the testimonials by Pandu Nayak (Chief Scientist Search) and Hyung-Jin Kim (VP Search) have been made available to the public. These are the most important core statements:
- ABC signals as the core: At the core, Google is based on the so-called ABC Signals (ranking factors): A stands for Anchors (left), B for Body (page content) and C for Clicks (user behaviour, more precisely: how long a user stays on a website before returning to the SERPs).
- “Hand crafted” vs LLMs: almost all ranking signals by Google are currently still “hand-crafted”, meaning that they are manually compiled by developers based on assumptions and tests. RankBrain and DeepRank are the only ones based on LLM technologies. The reason for this is traceability: If something doesn’t work, Google wants to understand why so they can fix the cause. With LLM systems this is possible either not at all or only through great difficulty.
- Q* as the trust metric: Q* (Q Star) is an internal metric that assesses the trustworthiness of a whole website (most often the domain) and is of significant importance for Google’s rankings.
- Navboost and click data: Navboost measures how often a specific group of users clicks on a specific hit in the search results. These values are based on the past 13 months and, at their core, represent a very large database.
- Confirmed leak and possible consequences: Google confirms that there was an informational leak regarding their ranking functions. However, from their point of view this leak is not sufficient to understand the functionality as a whole. Yet, if the court decides to oblige Google to pass on data concerning clicks, URLs and search requests to competitors, reverse engineering the Google algorithm could be possible.
