What Eric Lehman's Testimony Reveals
In September 2023, during the landmark DOJ antitrust trial against Google, a former Google software engineer named Eric Lehman took the stand. After 17 years working at the company on ranking algorithms and search quality, Lehman revealed something Google had long publicly denied: clicks influence rankings. The company had explicitly instructed him never to discuss this topic.
The testimony of Eric Lehman represents one of the most significant confirmations of Google's ranking practices to emerge from a legal proceeding. As a software engineer who spent nearly two decades working directly on how the algorithm determines website ranking and search quality, Lehman's statements carried particular weight. Crucially, he was testifying under oath--the legal requirement to tell "the truth, the whole truth, and nothing but the truth" added a dimension of accountability that Google's typical public relations statements never carry.
This revelation, combined with a major document leak in May 2024, has fundamentally changed our understanding of how Google's algorithm works. For SEO professionals, this confirmation carries significant implications for how we approach optimization strategies and communicate with clients about ranking factors.
The DOJ antitrust case, described as the biggest such legal challenge since the U.S. v. Microsoft case of the 1990s, sought to determine whether Google maintains an unlawful monopoly over the search engine market. Part of this investigation involved examining whether Google's practices give the company an unfair advantage, with Lehman called to testify specifically about the technical workings of Google's ranking algorithm.
Search Engine Land's coverage of the testimony became a focal point of the proceedings, revealing details that Google had long kept private.
The May 2024 Document Leak
In late May 2024, over 14,000 potential search ranking factors were revealed in leaked documents allegedly from Google's Content Warehouse API. This unprecedented leak offered marketers a rare glimpse into the inner workings of Google's ranking algorithm. While Google has since confirmed the authenticity of these documents, some information appears to contradict past statements from company representatives.
Click Metrics Revealed
The leaked documents identified several specific click metrics that Google uses as content relevance and user satisfaction signals. Our analysis identified four key metrics that SEO professionals should understand:
- goodClicks: This metric rewards search results that receive positive user interactions--when searchers click on a result and appear to find what they were looking for
- badClicks: Conversely, this metric demotes results that receive negative interactions--when users quickly return to the search results page after clicking
- lastLongestClicks: This measures the time users spend on a page before returning to the SERP, with longer engagement suggesting more satisfied users
- unsquashedClicks: This rewards clicks that are considered valuable and genuine, potentially filtering out automated or manipulative click patterns
Ovative Group's analysis of the document leak provides detailed breakdowns of these metrics and their implications for SEO strategy.
Understanding these SEO ranking factors is essential for developing an effective optimization strategy in light of this new information.
By the Numbers
17
Years Eric Lehman worked at Google
14,000++
Potential ranking factors revealed in leak
4
Key click metrics identified
Search Intent and Click Behavior
The relationship between clicks and rankings fundamentally comes down to intent matching. When users search for a query and click on a particular result, they're implicitly voting with their behavior. Google's algorithm interprets these votes as signals about which results best satisfy search intent for specific queries.
How Clicks Validate Content Relevance
The algorithm's use of click data serves as a continuous validation mechanism for content relevance. Traditional ranking factors like keywords, backlinks, and content quality provide static signals at the time of indexing. Click data, however, provides dynamic feedback about how actual users interact with search results in real-world scenarios.
When thousands or millions of users consistently click on certain results for specific queries, patterns emerge that reveal which content truly satisfies search intent. This is why understanding search intent has become such a critical component of modern SEO strategy.
Dwell Time as an Intent Signal
The "lastLongestClicks" metric is particularly significant because it measures not just whether users clicked, but how long they stayed before returning to the search results. This dwell time provides insight into content engagement levels. A user who spends several minutes on an article before returning to search may have found comprehensive answers, while a user who bounces immediately may signal that the content failed to meet expectations.
To improve dwell time and engagement, implementing effective web development best practices ensures your content is accessible, fast-loading, and provides an optimal user experience that keeps visitors engaged.
Technical Implementation of Click Tracking
Google's ability to track clicks stems from its position as the dominant search engine with billions of daily searches. Every search interaction, including which results users click on and how they behave afterward, generates data that feeds into understanding content performance.
The Role of RankBrain
Since 2016, Google has used its machine learning algorithm RankBrain as a core part of website ranking functions. While RankBrain does not use active user data in real-time, it does use historical search data as the basis within its functionality. This historical data includes aggregated click patterns that inform how the algorithm understands which results satisfy particular query types.
Machine Learning Systems: BERT and MUM
During Lehman's testimony, he discussed the differences between user data and training data in Google's machine learning systems. The company's advanced systems like BERT (Bidirectional Encoder Representations from Transformers) and MUM (Multitask Unified Model) rely on training data rather than active user data for their core functionality. Lehman testified that both systems are more important to website ranking than user data, as they can perform tasks with less user data incorporated into their algorithms.
By replacing user feedback with the unsupervised learning of raw text, these systems support Google's algorithm with reduced reliance on user data--though click data still plays a role in the broader ranking system. BluShark Digital's analysis provides additional context on how these systems interact with click signals.
For businesses looking to leverage AI for their content strategy, understanding how AI automation can help create content that aligns with these sophisticated algorithms is increasingly important.
The confirmation that clicks influence rankings has several practical implications for SEO strategy
Create Click-Worthy Content
Craft compelling title tags and meta descriptions that accurately represent what users will find. Ensure content delivers on the promise made in search listings.
Reduce Bad Clicks
Ensure content matches the search intent behind target keywords. When users click expecting one type of content and find another, they'll bounce quickly.
Optimize for Engagement
Focus on creating comprehensive, well-structured content that keeps users engaged through clear formatting, relevant internal links, and multimedia.
Match Search Intent
Understand the actual intent behind target keywords and create content that satisfies what searchers are genuinely looking for.
Measuring Click Performance
While Google doesn't provide click-through-rate data directly in Search Console for all queries, SEO professionals can monitor several indicators of click performance.
Available Metrics
- Search Console provides impressions and clicks data at the page and query level
- Position tracking tools can show CTR trends over time
- Analytics can reveal time-on-page and bounce rate patterns
Interpreting the Data
Compare CTR for your rankings against industry benchmarks for similar positions. If your CTR is significantly below average, it may indicate a mismatch between your search listings and user expectations--or that competing results are better satisfying search intent.
The Broader Algorithm Context
Click signals work in conjunction with numerous other ranking factors. The leaked documents revealed over 14,000 potential ranking factors, suggesting an incredibly complex algorithm. Click data represents one piece of a much larger puzzle that includes:
- Content quality and relevance signals
- Backlink profile and authority metrics
- Technical SEO factors like page speed and mobile-friendliness
- E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) signals
- Freshness and recency indicators
Throughout the leaked documents and testimony, a consistent theme emerges: user experience is paramount to Google's algorithm. Whether through click signals, quality assessors, or machine learning systems, Google's fundamental goal is showing users content that satisfies their search intent.
The confirmation that clicks influence rankings should reinforce what effective SEO has always prioritized: creating content that genuinely serves user needs. Rather than attempting to manipulate click behavior, focus on creating a comprehensive content strategy that addresses user questions thoroughly and keeps them engaged.