Takeaways from the Google Antitrust Lawsuit

On April 17^th, 2025, the US Department of Justice (DOJ) announced it had won its monopolization case against Google.

If this news sounds familiar, it’s because there were two nearly concurrent antitrust cases against Google: one about its search dominance and the other over its ad tech stack. The search trial ended with Judge Amit Mehta ruling against Google in September 2024, and remedies are now being considered. The second trial ended in April with US District Judge Leonie Brinkema ruling against the tech giant. Google says it will appeal both decisions.

What Happens Next with the Google Trials?

The remedy phase of the search antitrust trial started on April 21^st, and lawyers for the DOJ have taken the opportunity to warn that Google may try to further entrench itself in search through its use of artificial intelligence (AI). They are arguing that Google must be broken up under the monopoly laws and are calling for the following changes:

Forcing Google to sell its Chrome web browser
Eliminating exclusive deals with Apple and Samsung
Making Google share data with its competitors
Limiting its AI deals
Potentially selling Android

Of course, Google is arguing against these changes, claiming they would stifle innovation, harm user privacy, and give their Chinese competitors a technological edge.

As this portion of the trial is playing out, we’re seeing a treasure trove of information in the forms of documents and depositions that are providing insights into how Google uses data, trains its LLMs, and more. This episode has been the biggest set of “leaks” since the Google API leak last year.

Golden Nuggets from the DOJ v. Google Trial

Google used Search to train Gemini. This information shouldn’t be a surprise, but it was confirmed in a deposition. When Google first released the Search Generative Experience (SGE), it said it used search signals and user feedback to train the model. SGE eventually became Gemini, and now we’re seeing the next evolution roll out with AI Mode.

Navboost is not a machine-learning system. It is basically a huge table logging how many clicks a document (page) gets for a search query.

Google uses hand-crafted ranking signals. The humans at Google still play a larger role than machine learning in determining what ranks. When something in the algorithm doesn’t work as expected, engineers can more easily find and fix the problem compared to Microsoft’s more complex machine-learning systems.

The ABCs of ranking refer to Anchor links, the Body of the document, and Clicks. Anchor links refer to the source page pointing to a target page. The body of the document refers to the words in the document, and in this case, “clicks” means dwell time, or how the users interact with the page.

Google may give large language models (LLMs) a more prominent role in its search stack. It makes sense that Google would use language models to better interpret queries and summarize results, especially since we’re already seeing this in AI Overviews (and now AI Mode).

Click bias in ranking is a real thing that Google fights against. Two links of equal value may land in positions one and two, but people will naturally click on position one. So, Google made an adjustment to avoid that bias.

What Does All This Mean for SEO?

Many SEOs suspected much of this information was true, and now we have confirmation from the Google antitrust lawsuit proceedings. As we watch both trials’ remedies hearings and their results unfold, your ROI Revolution SEO team will continue to look for those nuggets of information that can help us navigate Google Search as it is today and how it may look in the future. In the meantime, the United Kingdom has filed a lawsuit of its own against Google for “near-total dominance” in search. It will be interesting to see how things play out across the pond.

If you have questions about optimizing for Google Search or how AI Overviews, Gemini, and AI Mode may affect your company’s visibility in the results pages, reach out to us to start the conversation.