User Engagement, Information Gain, and Google’s Search Brain: Insights from Bernard Huang (Ahrefs Evolve 2024)

Ever wonder why some content effortlessly climbs to the top of Google while others get lost on page two?

According to Bernard Huang during Ahrefs Evolve 2024, it all boils down to how Google learns from user engagement, using our clicks and interactions to separate the gems from the noise.

User Engagement Reinforces Learning

Google learns from us. Yes, every click, every scroll, every quick exit is a vote that tells Google what’s worth paying attention to.

There’s a loop here: results show up, we interact with them (or don’t), and Google learns which ones hit the mark.

It’s our engagement doing the heavy lifting. This process is a bit magical, but it’s actually our clicks that fuel the algorithm’s education, so to speak.

Key takeaway:

Google’s learning from us all the time. The more we interact with a result, the more it assumes it’s good stuff. It’s our clicks that fuel the algorithm’s education, so to speak.

Engagement and Information Gain Scores

Engagement drives something called the “Information Gain Score.”

Basically, when a search result gets a lot of interaction, it scores high in information gain, which is a fancy way of saying “this is probably useful.”

Less engagement? Smaller score. If people keep bouncing from a link, Google figures it’s not really giving anyone any new or valuable info, and down the rankings it goes.

Practical takeaway:

Engagement equals higher rankings. If you’re crafting content, aim for something users will actually find engaging and unique; otherwise, it’s going to sit somewhere no one will scroll to.

Information Gain: The Chicken or the Egg?

This raises an age-old dilemma: does Google’s Knowledge Graph drive user interaction, or does user interaction shape the Knowledge Graph?

It’s a cycle that’s hard to pin down. The “magic” here is the feedback loop — our interactions refine search results, which in turn refine the Knowledge Graph, creating an ever-evolving map of knowledge.

What this means for you:

Your engagement matters. Every click helps fine-tune Google’s “brain,” so it knows what’s actually relevant.

Think of it as a tiny vote in the grand scheme of online information.

How Google Uses Information Gain

Here’s a peek into Google’s process: it takes what users have seen on a topic, digs up some fresh but related content they haven’t seen yet, scores these new options, and serves up the top contenders.

This means Google’s always on the lookout for “new info” that expands on what’s already been consumed.

Why it matters:

Google is practically begging for fresh angles. Content that offers a new perspective or digs a bit deeper is gold.

Redundant info? Not so much.

What Users Think Actually Matters

User preferences drive everything.

Content that attracts higher engagement is viewed as more valuable and rises in search rankings.

Users clearly prefer content that feels authoritative and informative.

In Google’s eyes, high engagement equals valuable content, so it’s ranked higher.

For content creators:

Think quality over quantity. Engaging, useful content naturally pulls in users and keeps them around — good for them, and great for your rankings.

LLMs Produce Highly Probable (But Not Always Useful) Outputs

Large language models (LLMs) like OpenAI and Claude are excellent at generating content that feels “correct,” but they’re mostly just remixing what they’ve already seen.

The downside? They often don’t bring much new to the table.

If they’re trained on recycled SERP content, then it’s likely they’re regurgitating old ideas.

The bottom line:

LLMs are consensus machines, great for rehashing popular opinions.

But if you’re looking for fresh insights, don’t hold your breath.

LLMs Shine in Generating Consensus, Not Originality

LLMs tend to create a “consensus soup.”

Trained on mountains of search engine data, these models excel at summarising mainstream views.

The downside is obvious: they may make the web a bit stale, as everyone gets served the same, highly probable content.

For readers and creators:

If your content strategy involves these models, add your own spin.

Otherwise, you’re just another echo in the digital canyon.

Entity Overlap as a Measure of Originality

Google’s trick for spotting originality is called “entity overlap.”

Basically, if two pieces of content have too many overlapping “entities” (people, places, topics), Google marks it as redundant.

Fresh content has to mix in unique entities or provide new insights on existing ones.

Takeaway:

Google doesn’t just want reworded old news. It’s looking for genuinely original takes.

So don’t be afraid to go off the beaten path.

Low or No Information Gain = Redundant Content

Sites that offer the same information as others without adding any new context or value get ranked low in terms of information gain.

If every site repeats the same facts without adding context or value, it’s just clutter.

And clutter doesn’t rank well.

For website owners:

Find ways to stand out. If you’re just parroting what’s already out there, Google’s not going to reward it.

Aim for depth or a unique perspective.

Google’s Knowledge Graph

Google’s Knowledge Graph is like a web of interconnected topics — SEO, quality content, technical SEO, related entities — that helps it understand the relationships and context between topics.

This helps it serve up the best results for complex searches.

Real-world impact:

Google’s Knowledge Graph loves well-rounded content that covers related entities and subtopics.

If you want your content to rank, don’t be afraid to cover all angles and make those connections explicit.

Wrap-Up: What This Means for You

This guide is a wake-up call for content creators. User engagement and information gain aren’t just buzzwords — they’re the backbone of Google’s ranking system.

If you want to stand out in search, focus on engaging, original, and genuinely useful content.

And if you’re using AI tools to write, add a personal touch. Make it your own.

In short, Google wants fresh content, not echoes.

So give it something new to learn from, and watch your rankings climb.