Hi everyone, this is Neo.

Something big just happened in the SEO world — it’s like Google’s “nuclear codes” accidentally got leaked.

The trigger is the antitrust lawsuit between the US Department of Justice (DOJ) and Google. In the latest court filings, Liz Reid, Google’s head of search, had to lay her cards on the table to argue “we can’t share data with competitors” — explaining in detail exactly what core data Google owns and how critical it is.

For independent site operators and owners, this is a once-in-a-lifetime learning opportunity. Google never spells out its algorithm secrets this directly — but to win a lawsuit, this time they had to tell the truth.

Today, Neo is going to take you through a deep read of these documents: what’s the underlying logic of Google’s search ranking, and how should we adjust our strategies?

1. Why Does Google Refuse to Share Data at Any Cost?

In the antitrust trial, the court initially ruled that Google must provide its “user-side data” to competitors (like Bing, DuckDuckGo, etc.) to break the monopoly.

Google, of course, pushed back and immediately appealed. In the filings, Liz Reid stressed: these data are Google’s core assets — sharing them wouldn’t just let competitors “reap the rewards,” it would let spam flood the ecosystem.

This reveals two key pieces of information:

1. Google owns proprietary page quality and freshness signals.
2. User data is the fuel for training its AI models.

2. Uncovering Google’s Two “Secret Weapons”

The documents mention two core systems — concepts we could previously only guess at, now finally confirmed.

Secret Weapon 1: Google Glue

Think of this as Google’s “super-brain memory zone.”

Google Glue is a massive user-activity logging system. It doesn’t just record what users searched for (keywords) — more importantly, it records what users did:

• What results did the user see?
• Which link did the user click?
• How long did the user hover over a result?
• After clicking, how long did the user stay on the destination page?
• Did the user bounce back to the search results page (pogo-sticking)?

Neo’s take: Glue is like a tireless observer. It doesn’t care how many keywords you stuffed into your page — it only cares about real user reactions. If a user clicks your page and closes it a second later, Glue writes down a note: “this result isn’t trustworthy.”

Secret Weapon 2: RankEmbed BERT

This is the deep learning model behind Google Search.

The documents reveal that RankEmbed BERT is the core system used to “rerank” traditional ranking results. Its training data is hardcore:

• 70 days of real search logs (from Glue).
• Quality raters’ scores (human quality evaluators).

What does this mean? It means Google’s AI isn’t learning in a vacuum — it’s imitating human judgment. By learning from hundreds of millions of real clicks and interactions, it figures out what a “good result” looks like.

3. Every Page Has a “Spam Score”

Liz Reid also mentioned that every page in Google’s index is tagged with various annotations — including a spam score.

Google worries that if this data were made public, black-hat SEOs and spam site networks could reverse-engineer it, perfectly dodge spam detection, and pollute search results.

Neo’s take: For those of us running legitimate independent sites (white-hat SEO), this is actually good news. As long as we focus on delivering high-quality content and stay away from shady tricks, Google’s system can tell. Conversely, the people trying to game the system have nowhere to hide under Google’s “microscope.”

4. User Satisfaction Is the Only Truth

Reading through the whole document, the core logic is clear: user data is the lifeblood of Google’s ranking system.

In the past, our SEO mindset was more like “how do I make Google’s crawler understand me?” Now, we have to think “how do I make real users satisfied?”

The existence of RankEmbed BERT proves that even if your on-page SEO is flawless, if real users click in and have a bad experience (slow loading, thin content, can’t find the point), your ranking will still drop. Because AI adjusts rankings based on how users “vote with their feet.”

5. Practical Advice for Independent Site Operators

Now that we know these secrets, what should we actually do? Neo has summarized a few points:

(1) Extreme UX Is the Best SEO

Since the Glue system monitors every click and hover, your site experience has to pass the bar.

• Speed: your site must load fast — don’t make users wait.
• Navigation: let users find what they want at a glance.
• Mobile: make sure the browsing experience on phones is flawless.

(2) Content Should Be Direct and Solve Problems

Users search to solve problems. Can your article or product page deliver the answer on the first screen?

• B2B sites: skip the hollow company intro — show product specs, application scenarios, and customer cases directly.
• B2C sites: HD images, real reviews, and clear return policies increase how long users stay.

(3) Chase “Long Clicks,” Not Just Clicks

What we want isn’t “being clicked” — it’s “being read for a long time.”

• Embed videos on your pages; they significantly increase dwell time.
• Use rich multimedia elements and interactive design.
• Keep your internal linking structure sensible to guide users into reading more.

(4) Don’t Treat Users as Fools — and Don’t Treat Google as One Either

Since RankEmbed BERT learned from human quality raters’ standards, content expertise (E-E-A-T) is critical. Articles churned out by writers who don’t understand the industry can’t fool today’s AI.

Summary

This DOJ disclosure is actually a shot of confidence. It tells us: Google’s algorithm is becoming more and more like a real person.

As independent site operators, we don’t need to solve riddles or hunt for “black magic.” What we need is to return to the essence of business: serve every visitor well.

When users are satisfied with your site, Google will naturally hand you the traffic.

Hope this article gives you some inspiration for running your independent site.