The Shark Tank

In what is considered the biggest antitrust trial since the United States V.S. Microsoft case during the 1990s, Google has been placed under the legal microscope by the Department of Justice. This case alleges that Google has an unlawful monopoly over the search engine market and thus has an unfair advantage in the industry. As part of this case, several witnesses have been called to testify to the company’s actions and determine whether or not Google is acting as a monopoly.

One witness called to the stand is Eric Lehman, a former software engineer at Google. In his 17 years working for the company, he specifically worked on how the algorithm determines website ranking and search quality. What makes his testimony unique is that he was able to reveal classified information due to being under oath to tell the truth, the whole truth, and nothing but the truth. From his testimony, we are able to learn a lot about how Google determines ranking and the functions behind their machine learning systems.

What We Learned From the Testimony

In the testimony, Lehman touches on several topics related to how Google determines search quality and ranking. The company had expressly told him not to discuss the use of clicks in search and confirmed that, yes, click rates are a factor the algorithm uses when determining the ranking of a website.

Additionally, Lehman discusses the differences between user and training data. One of the concerns in this lawsuit is that Google uses user data as a core part of their algorithm, which would give the company an unfair advantage over others that do not have the amount of user data available at their disposal. Lehman states that their machine learning systems like BERT rely on training data rather than user data, which would not put Google at an unfair advantage over their competitors.

Speaking of machine learning systems, BERT and MUM were referenced several times when discussing user data. Specifically, Lehman testified that both systems are more important to website ranking than user data as they are able to perform the same tasks with less user data incorporated into their algorithms. By replacing user feedback with the unsupervised learning of raw text, these systems are able to support Google’s algorithm and its functionality with reduced reliance on user data.

However, there is one concern regarding user data that was mentioned during Lehman’s testimony. Since 2016, Google has ben using their machine learning algorithm RankBrain as a core part of their website ranking functions. While this algorithm does not use active user data, it does use historical search data as the basis within its functionality. By understanding how these algorithms work, we can better tailor our search engine optimization (SEO) strategies not only to what people are searching for, but for what these algorithms are prioritizing.

Improving Your Google Ranking

There are several SEO strategies that can be utilized to benefit your business’ digital footprint. However, knowing these strategies and how to implement them requires a high level of focused expertise in digital marketing. Our team at BluShark Digital is comprised of highly trained and experienced professionals that have honed their skills in website development, content creation, on- and off-page SEO, and pay-per-click (PPC) advertising. To see how our skill set can be applied to your business, contact us today!