Google AI Mode vs. Conventional Search & Different LLMs [Study]

Google launched its AI Mode within the US in Could, and SEOs at the moment are going through a brand new frontier.

On this examine, we got down to perceive how AI Mode compares to conventional Google search, AI Overviews, and two main LLM-powered search engines like google: ChatGPT and Perplexity.

Key Findings at a Look

• AI Mode prefers sidebar linking: 92% of AI Mode responses featured a sidebar displaying ~7 distinctive domains. These hyperlinks had a 51% area overlap and 32% URL overlap with Google’s high 10 search outcomes.
• Natural hyperlinks nonetheless matter: Within the 7% of AI Mode queries the place further hyperlinks appeared under the response, there was an 89% area overlap and 80% URL overlap with Google’s high 10 search outcomes.
• Reddit dominates: It appeared as a number one quotation supply throughout all LLMs we studied.
• Question intent impacts response size: Business queries triggered responses ~2x longer than informational ones.
• AI Mode seems extra just like ChatGPT than AI Overviews, indicating this new mode is extra than simply an upgraded model of AIOs. It’s an AI-powered search engine meant to compete with ChatGPT and different LLMs.

Methodology

What We Analyzed

To make sure balanced illustration, we chosen key phrases throughout 4 core intent varieties and two distinct search quantity tiers. Then, we queried them every into the 4 AI search platforms (desktop format) and recorded the outcomes.

• 5,000 randomly chosen key phrases from the Semrush database (just like the pattern utilized in our earlier comparability of ChatGPT Search, Bing, and Google)
• Over 150,000 distinctive citations throughout the platforms
• 4 equal teams of search intent: informational, transactional, business, navigational (evenly distributed)
• Two equal tiers of search quantity: 0-1,000 and 1,001-10,000 month-to-month searches
• 4 platforms: Google Search (high 10 natural rankings and AI Overviews), Google AI Mode, ChatGPT, Perplexity
• Overlaps between citations and rankings: When measuring overlap between Google and AI outcomes, we in contrast AI citations in every platform’s response to the highest 10 natural outcomes on Google for a similar question.

What We Regarded For

• Textual content and hyperlink overlap between platforms
• Response size variations
• Supply quotation patterns

AI Citations Solely Partially Overlap with Google Search

The primary query we requested was concerning the overlap:

How carefully did AI Mode and different LLMs mirror Google’s conventional high 10 rankings?

To measure this, we took every question and counted the variety of overlapping domains and URLs between every platform and Google’s high 10 natural rankings.

Whereas we noticed a strong overlap between Google’s high 10 natural rankings and LLM citations, the overlap was removed from absolute—and it different by platform.

• Perplexity confirmed the strongest alignment with Google’s high 10 rankings, with over 91% area overlap and 82% URL overlap. This implies Perplexity leaned closely on Google’s high 10 when selecting what to quote.
• Google AI Overviews additionally confirmed vital overlap—round 86% area and 67% URL—signaling that they nonetheless relied closely on Google’s conventional search index.
• Google AI Mode had a looser relationship: ~54% area and ~35% URL overlap. This urged extra unbiased retrieval conduct, particularly in sidebar outcomes.
• ChatGPT had the weakest overlap with Google’s high 10 rankings, with the lowest area and URL overlap. This echoed our earlier ChatGPT analysis that confirmed ChatGPT’s overlap was stronger with Bing than with Google. 

How AI Mode Compares to AI Overviews

Of their 2024 bulletins, Google framed AI Overviews as a method to assist customers get fast, AI-powered solutions—paving the best way for the totally interactive AI Mode launched only one yr later in Could 2025.

Whereas Google positioned AI Overviews as a bridge to its next-generation search expertise, we noticed just a few particular methods AI Mode strayed from the AI Overview playbook.

A Response + Sidebar with Hyperlinks Is the Default Format

In the event you’re not acquainted with AI Mode, right here’s what it seems like:

The most typical format we noticed included an AI-generated response and a sidebar with hyperlinks. In some circumstances, there was an area pack of outcomes. And in others, we noticed extra further hyperlinks under the response. 

Right here’s how usually we noticed every format:

• 92% of queries in AI Mode featured a sidebar with hyperlinks
• 7% of queries in AI Mode featured further hyperlinks under the AI-generated response
• 1.7% of queries in AI Mode included no hyperlinks in any respect
• 13.49% of queries in AI Mode included native packs
• This sidebar averaged seven distinctive domains (vs. simply three in AI Overviews)

Sidebar hyperlinks present Google experimenting with new sources, whereas hyperlinks positioned under the response stick carefully to its conventional natural rankings.

Whereas the hyperlinks under the response solely appeared 7% of the time, they’d a really sturdy overlap with conventional Google outcomes. 

This break up implies a twin system: AI Mode drew broadly for exploratory queries, however leaned on rankings for precision when wanted.

Particularly, we noticed AI Mode add further hyperlinks most frequently on navigational searches.

Navigational Queries Have been Most Much like Google Search Outcomes

We famous that the extra hyperlinks under the response had a excessive overlap with conventional search outcomes. However totally on navigational queries.

This is smart. In the event you’re utilizing Google to navigate someplace particular, you’d most likely like an inventory of hyperlinks (and never simply an AI-generated response). 

This might point out one distinction in how Google sees AI Mode and conventional search otherwise: AI Mode helps folks collect data, ask questions, and evaluate choices whereas they’re nonetheless making up their thoughts. 

Conventional search is there for when folks know the place they need to go.

AI Mode Exhibits Extra Distinctive Domains Than Different LLMs

One early critique of AI Mode is Google’s lack of transparency in the way it sends clicks and visitors to websites, very like with AI Overviews. 

Google lately confirmed that some AI Mode monitoring might be out there in Search Console, however these clicks might be lumped into total totals, not separated from conventional search information.

In our examine, we noticed that AI Mode confirmed extra distinctive domains than the opposite LLMs, on common, in its outcomes.

AI Overviews linked to a mean of three distinctive domains per question, whereas AI mode linked to 7. 

Perplexity and ChatGPT fell within the center.

AI Mode Most Resembled ChatGPT Search in Response Size

In terms of response size, we noticed Google’s AI Overviews stand out among the many LLMs because the briefest choice. 

AI Mode was closest to ChatGPT with response size, averaging round 300 phrases per response.

How AI Mode and Different LLMs Deal with Question Intent

Throughout all platforms, business and transactional queries triggered the longest and most detailed responses—usually double the size of informational responses.

We are able to see how a lot shorter AI Overviews are, on common. However extra attention-grabbing was the development of business and transaction responses being longer on each platform.

For SEOs, this implies aligning content material depth with question intent is crucial:

• Informational? Go for readability and conciseness.
• Business/Transactional? Increase, evaluate, clarify.
• For every kind of intent, match your content material depth to what AI instruments anticipate for every intent kind

A Notice on AI Search Intent: Whereas we used 4 traditional search intents for evaluation, AI-driven queries are more and more fluid. With conversational prompts, follow-up questions, and customized outcomes, AI-driven search is increasing the boundaries of conventional search intent. These 4 classes function a helpful reference level for Google search—however entrepreneurs ought to put together for a extra dynamic, multi-intent AI search panorama.

Quotation Patterns in AI Mode vs. Different LLMs

The primary graph on this examine confirmed the big-picture overlap between the LLMs and Google search. 

To search out extra patterns, we checked out just a few extra correlations throughout the platforms—evaluating what these AI programs cited with how content material ranked and carried out in traditional natural search.

Consumer-Generated Content material Platforms Are All over the place

We regarded on the most incessantly showing domains throughout the 4 LLMs (AI Mode, AI Overviews, ChatGPT, and Perplexity) and noticed a number of UGC and social websites on the high of this checklist.

• Reddit particularly dominated throughout AI Mode, AIOs, ChatGPT, and Perplexity
• In AI Mode, Reddit, YouTube, and Fb appeared in 68%+ of outcomes with further hyperlinks—outpacing conventional model web sites
• Mapbox.com and openstreetmap.org, two platforms for constructing customized interactive maps, cracked the highest 10 regardless of zero visibility on Google for a similar queries

Google’s AI cope with Reddit certainly had an affect on its standing right here, and Reddit’s visibility will doubtless proceed with extra AI Mode adoption. 

However why did these map websites see this enormous visibility in LLMs regardless of not seeing the identical outcomes on Google search? 

One motive could possibly be Mapbox’s award-winning product documentation.

With clear, structured, and developer-focused content material, Mapbox gives content material that’s simple to reference.

OpenStreetMap (OSM) is an open-source mapping challenge. It is a go-to reference for geographic information—from avenue names to pure landmarks—as a result of it’s:

• Repeatedly up to date by contributors around the globe (just like UGC)
• Out there below a permissive license (Open Database License)
• Steadily utilized in each educational and business functions

Implications for Entrepreneurs:

• Now’s the time to develop content material methods for UGC platforms, not simply your web site
• Monitor and have interaction with discussions about your model/business on Reddit
• Create thorough and useful documentation for those who supply a technical services or products
• Take into account YouTube as a major content material distribution channel

Sturdy Rankings on the Area Stage Correlate with AI Visibility

One of many clearest alerts in our examine: domains that ranked nicely in Google’s natural high 10 outcomes have been way more prone to be cited in AI-generated solutions.

The connection was constant throughout all LLMs we examined. As a site’s rely of high 10 rankings elevated, so did its presence in AI citations—with remarkably sturdy correlation on the area stage.

What this tells us is that throughout each channel, in case your area will get plenty of high 10 Google rankings, it is very doubtless that you simply’ll get plenty of AI citations as nicely.

Whereas correlation doesn’t affirm causation, the power of the connection between natural rankings and AI citations means that dominant domains in conventional search will proceed to steer in AI visibility.

Rating nicely for extra particular, related queries may give LLMs extra choices to select from when selecting citations.

RAG-based fashions like Google’s AI Mode begin their course of by retrieving top-ranked outcomes earlier than augmenting and producing their last reply (RAG = retrieval augmented technology). 

That’s one motive why high-ranking domains have been extra prone to be cited.

Why the Area Correlations Are Stronger than the URL Overlap Between LLMs and Google

LLMs like ChatGPT present a powerful correlation with domains that rank nicely in Google’s natural outcomes—however that doesn’t imply they cite the identical URLs you see within the high 10.

As a substitute, LLMs usually pull from completely different pages inside the similar trusted domains. That’s why we see excessive domain-level correlation however much less direct overlap on the URL stage.

Moreover, as a result of AI Mode might tailor its responses based mostly on consumer context—corresponding to search historical past or location—citations might differ barely from one particular person to a different.

Conventional search usually highlights homepage-style or pillar content material. 

LLMs, particularly when answering nuanced or refined queries, have a tendency to achieve deeper—citing subpages, particular weblog posts, or assist articles that higher match the question context.

In reality, our current analysis on the influence of AI on search engine optimization visitors discovered that almost all of ChatGPT’s citations in that examine pattern have been pulled from URLs rating past place 21+ on Google.

The takeaway: LLMs might belief the identical sources, however they received’t all the time pull from the identical web page.

What AI Overviews and AI Mode Inform Us About Google’s AI Technique

Google’s AI Overviews served as a testing floor for its full AI Mode expertise. 

Present state:

• AI Overviews at present seem for ~15% of key phrases in keeping with Semrush Sensor
• AI Mode natural outcomes confirmed a 58% URL intersection and 88% area intersection with AI Overviews

The excessive area overlap (88%) suggests constant belief alerts throughout each codecs, whereas the decrease URL overlap (58%) hints at evolving retrieval logic in AI Mode.

In the event you’re already being cited in AIOs, you are heading in the right direction—however AI Mode might introduce extra variation in which pages it pulls from, even inside the similar trusted domains.

Sensible Takeaways for search engine optimization & Advertising Groups

Trying on the information is nice, however what must you do subsequent? 

Based mostly on the findings, right here’s an inventory of steps you may take to place your self now for what’s coming forward.

1. Audit your AIO presence with Semrush Natural Analysis
2. Audit your Reddit technique. It is unavoidable throughout all AI platforms.
3. Don’t ignore conventional rankings. They’re nonetheless a gateway to AI visibility, however they received’t be the visitors drivers they was. Monitor conventional rankings (on Google and Bing) alongside ChatGPT with Place Monitoring.
4. Create intent-specific content material lengths. Match AI expectations for every question kind by offering loads of particulars for business and transactional subjects.
5. Monitor AI citations. Monitor how and the place your model seems throughout AI platforms (use Semrush AI Toolkit for this).

Trying Out on the AI Search Panorama

AI search is not changing conventional search engine optimization, nevertheless it’s creating its personal class. 

This rising self-discipline, generative engine optimization (GEO), amplifies the significance of sturdy fundamentals whereas including new distribution channels and methods to grasp. 

LLM visibility will doubtless come not simply from rating excessive, however from being reference-worthy, reliable, and cited throughout platforms the place customers make selections.

Key predictions based mostly on our information:

• Conventional search engine optimization fundamentals will stay essential for AI visibility
• Consumer-generated content material platforms will grow to be much more essential
• AI instruments will proceed creating distinct quotation personalities
• Generative engine optimization will grow to be a necessary layer in trendy content material methods