Let’s open the hood here and try to better understand how their AI Overview system works, how it might be connected to other parts of their eco-system, and what the future might look like if Google keeps going in this direction.

Google filed for the patent in 2023, not earlier - potentially indicating that they were caught off guard by the success of ChatGPT. The patent was written by several Google research scientists and engineering team members from the search team. Many of them have participated in other Google patents, and are, obviously, intricately familiar with how all of the Google systems are intertwined.

Under the hood of Google’s AI Overviews (f.k.a. SGE)

The patent itself is quite long (30 pages when printed!) so i’ll spare you all of the tech jargon and give you a layman’s overview of how it works…

Google starts with its typical query ingestion and understanding process. Once has determined the context of the query, things begin to rapidly diverge from the standard way Google determines search rank.

Using the query, and other information it feels is relevant, Google retrieves several SRDs or Search Ranking Documents, which are pages that could potentially rank for a given query. These are returned and ranked based on several factors - some of the usual suspects like trustworthiness, authorship, inbound links, etc. They also consider how “summarizable” a given article is based on several internal factors not listed in the patent.

This is likely why you don’t always only see top-ranking documents as sources in SGE.

From there Google selects which of its LLMs it might want to use based on which documents it selects. We know Google has several different models to choose from when considering how to generate a response. They can use as many as they deem necessary at this step.

Once snippet responses are generated, they are recombined using a separate engine that brings all of the responses together in a sensible way. Think of this like taking pieces from different Lego sets and building a new thing.

At that point Google begins checking the response for accuracy and measuring the confidence level. This is done one sentence at a time using the cosine similarity of the original source snippet to the Google-generated snippet.

The patent specifically states that higher confidence responses have a lower chance of having their sources cited. To me, this feels similar to what they do with their Instant Answers featured snippet.

Once each snippet in a generated response has been verified with at least one SRD, the complete response is sent back to the client device for rendering.

Important to note is that each generated snippet has “meta-data” which includes what SRDs it was generated from, as well as what SRDs it was verified against. We only get to see a small portion of that on the front end in the format of clickable links in a sliding panel under the generated result.

How is information verified and cited according to the patent?

Here is the portion of the patent that describes their process for verifying and citing information:

Once an answer is generated, Google begins the process of verifying the individual sentences within the response one at a time.

They select one of the possible search ranking documents associated with generating the answer or ranking organically for a given query. They do this by extracting a portion of the SRD and comparing the content embedding to the content embedding of the LLM-generated response. The passage is then verified if the distance between the two embeddings is less than a specific threshold.

Once a section is verified, the associated url that was used to validate the portion and the verified AI output are “linkified” or connected together. In the example below, the LLM-generated response about “Bad mood may cause…” was verified in the linked Travel China Guide article listed below.

Once that source is clicked, you are taken directly to that portion of the article with the content used to validate the LLM-generated response, highlighted. If one wished (cough Google cough), incorrect answers could be debugged in this way to determine patterns and improve the accuracy of the generated responses.

What do the different colored backgrounds of AI Overviews mean?

SGE was the first real introduction of color to the Google SERP page - but what do the different colors mean?

Colors seem to illustrate Google’s level of confidence in the synthesized answer that its system provides. They do reference this in the patent but only using general examples.

Here is how the use of color is described within the patent:

How should we interpret this? Well in my 24 hours of testing post-Google I/O, using a logged-in, but non-Search Labs enabled account, I was able to get 5 different color variations of the AI Overview panel.

This feels like an extrapolated version of what Gemini does in its chat interface by highlighting parts of the text it has found content similar to on the web in green, and highlighting content it has found various answers for in orange.

With that, we can possibly deduce that the scale likely goes from Green → Orange in terms of how confident Google is in its generated response.

In the example above, the answer to when is the Lunar New Year, or where do puffins live, do have more fixed answers than something like how to make a latte, or who is the richest person in the world.

The patent does describe what is called “confidence annotations,” perhaps color is how Google chose to implement this. Either way, it would be good for them to communicate this as most of the medical queries I found with AI Overviews…were red.

Why do we see things that don’t make sense in certain responses?

Besides the fact that LLMs have hallucinations, Google’s process for validating the accuracy of a generated passage using cosine similarity sits on a bit of shaky ground.

Researchers from Netflix observed that blindly using cosine similarity for semantic similarity between learned embeddings can generate meaningless results.

We can assume that Google is just as good, if not better, than Netflix at building and comparing embeddings, but this study does prove there is at least some wiggle room when using this method to compare snippets of text.

Especially when not manually reviewed, and trained by a human, like their search results have been using quality raters.

Additionally, source documents selected for summarization are not selected in the same way as Google’s organic algorithm works. The patent specifically calls this out by stating that Google may use one or more ranking systems to determine which documents it uses as sources for LLM summaries. See below:

Since this is all being done algorithmically, they can select a source document that meets the criteria of the selection engine but is factually inaccurate.

That is the case in the below example suggesting that it is ok to drive while drunk in the country Barbados shared by MartinMacDonald on X:

In this case, and in a few others, it would seem that the SRD selection process needs to be refined, or that the validation engine needs to seek out and confirm that a generated passage is similar to passages found on multiple high-authority websites before validating the response in the response confidence engine.

Unused parts of the patent and how this experience might evolve

Aside from just making this product work correctly so they don’t destroy the goodwill they have built up over 15 years (maybe they already have) - the patent outlines a few additional behaviors not currently seen in the production version of AI Overviews.

Real-time answer refinement

The patent also describes a way for this AI Overview panel to adapt based on user behavior. This behavior might include clicking one of the sources, toggling one of the listed source panels, or perhaps even click/hover behavior like they use for standard organic ranking.

In practice this seems more like how a pack of PAA questions work, in that when you toggle one open, you’re often presented with several more related questions below the original 4.

This makes sense as one of the inventors of the SGE patent was also the search team engineer behind the implementation of People Also Ask several years ago.

Integration of user profile data

The true moat here for Google is to leverage its understanding about you as a human and then generate an answer based on your past searches, as it understands you have a different potential level of understanding about a topic.

The patent does indeed discuss this:

While this likely needs more consideration due to all of the privacy issues it brings up - it give Google a big leg up over its rivals and really shows the direction this could all be going in a few years.

Imagine if you were a dog owner researching breeds, vets, or pet accessories for hard chewers. Let’s say at some point you also searched for a brush to comb through a german shepherd’s coarse hair. 6 months later you are searching for something like “Why is my dog limping?”

The magical part would be if Google can remember that not only do you have a German Shepard, but also knows they are more likely to have hip dysplasia. Google then generates an AI response stating that it is common for shepards to have hip dysplasia and suggests you visit your local vet with the specific phone number you’ve searched for before.

Magic.

This is also likely why it currently requires a logged-in profile to display a LLM-generated overview.

Wrapping up

Who knows where this is going but it is certainly a very exciting time to be alive.

We cannot go back to how it was before, the cat is out of the bag. Whether we like it or not, information is beginning to become a commodity, and standing out is going to be increasingly hard if you are in the content business.

The SEO community would be wise to begin diversifying and researching new ways to think about their craft.

As ancient Greek philosopher Heraclitus said - the only constant, is change.

If anyone else has spent time digging into this, let me know in the comments!

Now, I know that might sound like heresy to some SEO professionals who have built entire careers around the ever-evolving tactics and strategies of search engine optimization. But hear me out…

At our core, we SEOs are analytical problem-solvers, constantly seeking answers to questions from clients, trying to identify which ranking factors matter most, and digging into data to confirm we’re on the right track. We’re taught from an early age about cause-and-effect, and it’s our job to deduce what actions will drive more organic traffic.

This leads us to gravitate towards strategies employed by other smart people in our field. We study data patterns and trends, hoping to gain an edge and squeeze out additional growth. This approach isn’t unique to SEO – nearly all marketing disciplines rely on analyzing what works and optimizing from there.

However, when it comes to SEO specifically, this tunnel vision on tactics often causes us to miss the bigger picture…

The reality is that while the specific methods and “tricks” may change over time, the fundamental principles behind effective SEO have remained relatively constant. If you thought long and hard about what users want, put yourselves in their shoes, served your content in a quick and non-disruptive way, you were just as likely to rank 10 years ago as you are now. None of that has changed since we started.

At the end of the day, your website must legitimately earn its rankings – the content is a part of it, but not the only part. The several years of “content is king” served as sort of a red herring here and I believe did a bit of a disservice to the community. With user behaviors evolving to be increasingly impatient and judgemental thanks to social media’s influence, we need to lean into those harsh realities rather than pretending they don’t exist.

Google has continued serving as the shepherd for the internet, helping propel it forward in usability, security, and overall user experience. To align with Google, your users, and elevate above all of this chaos we need to think more like Warren Buffet SEOs than Roaring Kitty SEOs…value is the name of the game.

Google has served as the lighthouse of the internet

For over 15 years, Google has acted as the navigator – the guiding light allowing users to traverse the vastness of the internet. As the web and mobile adoption exploded, so did the volume of content being created. This necessitated a curator to maintain accessibility and quality control over this burgeoning digital library.

Google stepped up as that modern-day librarian. Not just making information discoverable, but setting standards to preserve the internet’s integrity. Major algorithm updates pushing for SSL, mobile-first indexing, Panda, Penguin, Medic, CWV all underscore Google’s commitment to refining search results for a better user experience if you step back and look at the forest, not just the trees.

These updates targeted low-quality content, untrustworthy sources, and misinformation – particularly in sensitive categories like health. The goal was to ensure users found reliable, relevant information. And while the details changed with each new update, the underlying principle remained constant – improving the overall quality of how people experienced the internet.

It’s also important to understand that Google itself was going through growing pains during this time. As a commercial entity, it had to carefully balance improving the organic search experience with protecting its advertising revenue stream. The world’s most popular search engine faced constant criticism about the role of paid ads overtaking or influencing organic listings.

This delicate balance sometimes led to adjustments that appeared user-unfriendly in pursuit of additional ad inventory. However, long-term, Google seemed to recognize that providing the best overall user experience – both organic and paid – was paramount to maintaining its dominance.

Yes, Google introduced new ranking factors and guidelines along the way. But these were more like “nudges” recommending best practices, not literal checklists for guaranteed rankings. Too often, SEOs got distracted by the latest tactic or technical trick being hyped on social media, promising overnight success if implemented.

But the truth is, while tactics evolved, Google’s strategic direction centered on constantly raising the bar and pulling the entire internet experience forward for users. The specific methods changed, but the fundamental drive toward a more accessible, credible, and enjoyable digital world remained resolute.

So rather than frantically chasing every new trend, we’d be better served by reorienting around Google’s core mission as the steward relentlessly dragging the internet forward.

Rethinking Blame: Why Your Content Must Earn Its Rank

While criticism of Google’s search results periodically surfaces, often lamenting the difficulty of finding quality information or accusing bias against smaller sites, this viewpoint misses a critical point: the onus isn’t solely on Google, but also on content creators truly understanding and adhering to the principles of effective SEO.

Google’s mission has always centered on providing users with the exact information they need, so they keep returning to Google as their primary source. While ad revenue is a factor, the foundation is delivering valuable, engaging search results. This underscores that simply having content doesn’t guarantee success – that content must resonate with searchers’ needs.

Understanding those needs requires going beyond just matching keywords to intent. Search intent exists on a spectrum, not as black-and-white categories. The same query could have informational, commercial, or other underlying motives depending on the specific user and context.

Earning high rankings demands anticipating not just why the user searched initially, but what they’ll want to do after consuming your content. What are their likely follow-up questions or next steps? The most successful content acts as the nexus for that entire exploratory journey.

It’s a strategic, empathetic mindset shift: deeply understanding the specific users stumbling across your content, their perspective, motives, future queries, and desire paths before they even realize it themselves. You could be the foremost authority, but without this intentional empathy for the full user experience, your content’s ranking potential will be limited. Short-term isolated success is possible, but the goal should be sustainable authority.

Google’s algorithm will never be perfect, and we should never expect it to be. It has to meet the needs of everyone collectively and you the individual all at the same time. However, as SEOs, our role is to fundamentally adapt by understanding how ranking systems evolve, but more importantly, user intent evolves, ensuring our content doesn’t just exist, but thrives by serving the entire range of user needs.

Above the fold matters more than you think

Google heavily relies on user behavior and engagement signals to gauge how helpful a piece of content is for a given query. While their algorithms have evolved in complexity, this foundational aspect hasn’t changed – first impressions and initial engagement are crucial.

It’s akin to walking into a disorganized, off-putting retail store – you’re likely to quickly turn around and leave. The same applies to websites. The best SEOs I know think about user experience and design just as much as they think about content optimization and technical tricks. 

The goal is to pique your users’ interest immediately and then to graciously lead them down the rest of your page experience.

Too many sites bury the lede, attempting to maximize ad impressions or force reading the entire article. But this directly conflicts with today’s impatient, social media-trained user base. Who has time for long intros or buried intent anymore?

Over the past few months, there have been a handful of “exposes” talking about how “unfair” Google is. Lots of sites that exist for the sake of driving affiliate clicks (but wearing the clothes of a site providing unique value, are seeing their traffic decimated.

To captivate interest from the second users land on your page – the pivotal “above the fold” moment – you need to leverage some cognitive biases:

The Sunk Cost Fallacy suggests people are more likely to continue an endeavor after investing in early effort. By delivering an engaging hook or fascinating fact immediately, you not only earn crucial second-look interest but also make users more inclined to continue through your content because they’ve already begun investing attention.

The Halo Effect means larger, established brands tend to get the benefit of the doubt for quality and credibility over lesser-known entities. Subconsciously, familiar brands are more trustworthy sources, so they earn extra consideration. This bias partly explains why major publishers often outrank smaller sites, regardless of content quality. If you don’t have a huge brand yet leverage the power of other well-known brands by adopting a badging system, featuring logo links to other places your content has been linked from, and more. There are ways around this particular issue – but that is for a future article.

To maximize visibility and engagement, optimizing for these cognitive biases is critical from the moment users arrive. An impactful above-the-fold experience doesn’t just convince them to stick around – it subconsciously shapes their perception of your content’s credibility and comprehensiveness before they’ve read a single word.

Immediate engagement or a particularly interesting fact or tidbit above the fold is what you’re looking for. Not only will that person have the momentum to keep engaging with your page, but it increases the chance that the user will engage with something else down below because they have already engaged with something on your page.

Rethinking the “Shopping Mall” Analogy

Google is often compared to a vast digital shopping mall, with websites as the storefronts vying for attention. But this framing is usually misguided. Let’s explore it with a fresh perspective:

Imagine searching for a new pair of basketball shoes. You wander the mall, drawn to both big-name retailers like Foot Locker as well as lesser-known indie stores like ShoeExperts and BestShoes4U.

You enter ShoeExperts and are met with a bland, lackluster presentation. The disorganized array of shoes and distant (albeit knowledgeable) staff fail to captivate you, so you promptly leave to resume your search.

In contrast, Foot Locker greets you with an immersive basketball court atmosphere – theming, creative product displays, and associates in referee shirts. You’re enveloped in an engaging brand experience that compels you to browse and purchase.

This highlights two critical points often missed in the mall analogy:

1. Google’s role isn’t as a curator judging the inherent value of each “store.” It’s a neutral hosting platform, with the onus on individual sites to present their “merchandise” (content) engagingly.

2. Larger brands often have an innate advantage not due to favoritism, but because of psychologically appealing to humans’ familiarity biases and ability to create compelling browsing experiences.

The success differential between ShoeExperts and Foot Locker doesn’t stem from the mall’s shortcomings, but from the latter investing in presenting offerings persuasively. While ShoeExperts may possess richer domain expertise, their bland packaging fails to captivate buyers.

This underscores that simply having superior information isn’t enough. To win in the Google “mall,” you must optimize not just for providing value, but making that value accessible, visible, and intriguing from the moment a user first encounters your “storefront.”

But the core premise holds: smaller publishers often err in battling major brands head-on for high-volume keywords, when a wiser approach is first cultivating visibility and engagement with more niche audiences before attempting to swim upstream.

An Evolved Mindset for SEO Success

After unpacking all these principles, one core truth emerges: SEO has never fundamentally changed because Google’s mission has always remained constant – providing the best possible user experience.

We often get caught up in the wrong notion of quality, fixating solely on content attributes like word count, media, or technical specifications. However the content is just one part of the overall user experience. It’s the seamless combination of engaging presentation, anticipating searcher needs, and sustaining an intuitive journey that truly denotes quality in Google’s eyes.

Search rankings have been and will always be dictated by organic user behavior signals – the clicks, hovers, pogo-sticking, and engagement metrics that feed Google’s increasingly intelligent algorithms. While AI and machine learning allow more nuanced content comprehension, those models still rely on user data as the ground truth for what constitutes quality.

So as SEOs, we must evolve beyond just checking boxes for content creators. We need to become experienced architects and storytellers – experts at not just conveying information, but crafting intentional, visually-captivating digital experiences that compel users down an intuitive path toward fulfilling their needs, however ambitious or tangential.

The best SEOs aren’t just coaches or consultants, but psychologists who deeply understand how to spark interest from the very first impression, leveraging core cognitive biases like the Sunk Cost Fallacy and Halo Effect to their advantage. It’s about optimizing for attention and emotion as much as keywords and semantics.

In this increasingly crowded digital landscape, resonating with users will require rethinking our approach from the ground up – moving our craft away from myopic tactics and toward a holistic, empathetic discipline of experience optimization.

Mastering SEO was never about gaming the latest algorithm update. It’s always been about building a connection with your audience, earning their trust, engaging their curiosities, and serving their needs better than anyone else can. That fundamental truth remains immutable, even as the specific methods and technologies evolve.