How does re-ranking work on Google?
Posted: Sun Apr 20, 2025 10:02 am
In his session "How Google Reranks," Hanns Kronenberg ( Chefkoch ) used a great example to demonstrate how one can gain insights into Google's work through one's own web projects: In 2021, the Chefkoch team launched their own project to improve internal search. According to Hanns, this process was one of the most "insightful" in his time as an SEO.
At the beginning of the session, he asked what the appropriate ranking factors would be for a belize phone number data search query for "strawberry cake." The SEOs present were asked to suggest factors – and promptly named the usual suspects, such as title, rating, age, and downloads.
But then Hanns explained an important problem: How do I ensure that the search results that follow these criteria actually satisfy users? Chefkoch has integrated an AI-supported re-ranking after the initial algorithmic ranking. This takes user signals from previous searches (click-through rate, etc.) into account. And lo and behold: Chefkoch and Google arrive at the same result for position 1 for the topic "strawberry cake."
Thus, this small experiment already revealed the impact of user signals in Google search. Using the example of "green sauce," Hanns then demonstrated which criteria play a role alongside the hard algorithmic factors and how the clickability of recipe images affects the results for the topic "strawberry cake": For example, a recipe could be boosted from 3 to 1 overnight by a more attractive image.
At Google, Twiddlers are responsible for reranking search engines—an important topic for all SEOs. Required reading in this context: the document on Twiddlers, which was made public during the antitrust proceedings against Google.
In summary, Hanns described the re-ranking as a Formula 1 race: There's a qualifying session in which everyone fights for the best starting position. This is followed by the actual race, in which the participants compete against each other—and this race is the re-ranking of the search results.
At the beginning of the session, he asked what the appropriate ranking factors would be for a belize phone number data search query for "strawberry cake." The SEOs present were asked to suggest factors – and promptly named the usual suspects, such as title, rating, age, and downloads.
But then Hanns explained an important problem: How do I ensure that the search results that follow these criteria actually satisfy users? Chefkoch has integrated an AI-supported re-ranking after the initial algorithmic ranking. This takes user signals from previous searches (click-through rate, etc.) into account. And lo and behold: Chefkoch and Google arrive at the same result for position 1 for the topic "strawberry cake."
Thus, this small experiment already revealed the impact of user signals in Google search. Using the example of "green sauce," Hanns then demonstrated which criteria play a role alongside the hard algorithmic factors and how the clickability of recipe images affects the results for the topic "strawberry cake": For example, a recipe could be boosted from 3 to 1 overnight by a more attractive image.
At Google, Twiddlers are responsible for reranking search engines—an important topic for all SEOs. Required reading in this context: the document on Twiddlers, which was made public during the antitrust proceedings against Google.
In summary, Hanns described the re-ranking as a Formula 1 race: There's a qualifying session in which everyone fights for the best starting position. This is followed by the actual race, in which the participants compete against each other—and this race is the re-ranking of the search results.