The data for this study was collected using STAT — you can see the full data here.
Google’s AI Mode is introducing new patterns in how search results are assembled. While traditional rankings still matter, it’s not the only route to organic visibility.
This report analyzes nearly 40,000 search queries to uncover the biggest winners of AI Mode citations and how often overlap occurs with traditional SERPs.
Key takeaways
- 88% of AI Mode citations are not in the organic SERP for the exact match query
- YouTube is the second most cited external source in AI Mode
- 10% of citations come from the top 4 external sources
- 96% of responses include at least one citation, with most pulling from 10 or more unique URLs.
Research methodology
To understand how AI Mode generates results and selects citations, we analyzed a structured dataset that included nearly 40,000 queries across desktop and mobile in the US and the UK.
Each AI Mode query returns up to 38 results, with data on type (paragraph, list, etc.), traits (citation, answer, etc.), position, and layout structure.
Traditional SERPs for the keywords were extracted to measure overlap with AI Mode citations.
88% of AI Mode citations are not in the organic SERP for the exact query
There’s not a lot of overlap between AI Mode citations and Google’s top 10 results for the same query — just as we’d expect given the fan out methodology.
For this study, we looked at two types of overlap:
- Strict URL match: The exact page appears in both AI Mode citations and Google SERP
- Site-level match: The same domain appears, even if it’s a different page
Only 1 in 10 AI citations match the exact URLs in Google’s top 10 organic results. Even when compared at the domain level, just 1 in 5 citations come from the same websites that appear in the top 10 for that query. When overlap happens, it’s concentrated near the top, but still far from guaranteed.
As Tom Capper explains:
“AI Mode is branching out to a broader set of queries and topics rather than just the exact one you typed in... It makes these sort of ‘fan-out’ queries and aggregates results.”
In simple terms, AI Mode runs multiple related searches behind the scenes (e.g., variations, subtopics, adjacent intents). It then aggregates citations from those related queries, not just from the top results of the original query.
Implications for AI SEO strategy
Low overlap is a reflection of how AI Mode is designed
While strict overlap is only 12%, it’s not a flaw. Instead, it’s evidence that AI Mode is:
- Drawing from a wider knowledge set
- Favoring authority across related topics
- Surfacing sources that serve the full context
Ranking in the top 10 helps, but requires broad keyword coverage to guarantee citations
In this dataset, a higher Google rank clearly increases the odds of being cited in AI Mode, but even the #1 result is only cited some of the time.
For most queries, while there is some shared sourcing between AI Mode and organic results, the citation list is mostly broader than the SERP because AI Mode uses a wider set of queries.
10% of citations come from the top 4 domains
The majority of AI Mode answers pull from a small set of highly trusted domains, probably due to their prominent positions in the top few organic results for the fan-out queries.
In the table below, we analyzed:
- % of responses: How often a domain appears at least once in an AI Mode answer
- % of citations: The domain’s share of total citation volume across all responses
The top four domains account for 10% of all citations, signalling the sources AI Mode prefers when generating responses.
Who are the biggest winners of AI Mode citations?
- Reference authorities (Wikipedia): No surprise, Wikipedia is the most cited source by far and used as a factual anchor across categories.
- Google-owned properties: YouTube and Play Store links dominate. Proof that Google’s AI Mode favors its own ecosystem.
- Platform truth (Reddit, Facebook, and LinkedIn): The model relies on UGC platforms for context, community input, and social proof.
- Institutional medical authorities: Mayo Clinic and Cleveland Clinic are recognized thought leaders, suggesting a strong trust signal in the health vertical.
Implications for AI SEO strategy
Brand presence off-site matters
Expand your citation surface area. UGC domains like Reddit, YouTube, Facebook, and LinkedIn appear frequently. If your brand is absent or poorly represented on those ecosystems, you may lose mindshare even if your site ranks.
Video content is no longer optional
Include video production in your content strategy. YouTube ranks second among the most cited domains in AI Mode. That means high-quality video content is a core part of how the model builds trustworthy answers.
96% of responses include citations, usually 10 or more
AI Mode is citation-heavy by design. Out of nearly 40,000 responses analyzed, 96% included at least one citation.
This clearly contradicts the assumption that AI Mode skips attribution. On the contrary, it’s citation-heavy, often pulling multiple URLs into the summary even when it looks minimal on the surface.
Implications for AI SEO strategy
Treat AI Mode citations like a bibliography
With 91% of responses citing 10 or more sources, AI Mode behaves more like a bibliography than a top‑10 list. However, the goal isn’t just to get your site cited, but to earn a mention in the paragraphs that appear on already-cited domains. Focus on being part of the answer, wherever it appears.
Avoid generic content
Pages that get cited tend to cover a specific micro-intent. Avoid generic overviews and ensure your pages focus on a single, well-defined intent.
Design for extractability
According to John Iwuozor, “AI Mode processes content at the passage level, pulling out modular chunks to build its responses.”
Since AI Mode is pulling text-based citations, structure your content with clear subheadings and focused paragraphs. Make it easy for the model to identify what your page answers and where it fits in the bigger picture.
What SEOs should do next?
1. Cover the whole topic, not just the head term
AI Mode doesn’t rely on a single search result. As Tom Capper explains, it runs multiple related queries behind the scenes and builds answers from a broader intent set.
That means a high-ranking page isn’t always enough. Content that covers adjacent questions, subtopics, and follow-up queries has a better chance of earning citations even if those pages don’t rank at the top.
The best way to do this is through topic clustering. Create focused pages that own specific angles of a broader theme to increase your chances of getting mentioned in AI Mode.
2. Off-site visibility is essential
UGC platforms like Reddit, YouTube, Facebook, and LinkedIn account for a meaningful chunk of citations.
If your brand isn't on these platforms, you risk losing visibility in AI Mode, even if your website ranks well.
To improve off-site visibility:
- Contribute to expert-led threads on Reddit and Quora, especially for niche or technical topics.
- Build consistent third-party profiles on platforms like Yelp or TripAdvisor if you operate in a local or service-based industry.
- Leverage partnerships and guest content to earn citations from authoritative domains.
- Get active on social: Post thought leadership content on platforms where your audience spends time, like LinkedIn and Bluesky.
3. Video content plays a central role
YouTube is the second-most-cited domain in AI Mode, ahead of Reddit, Facebook, and LinkedIn.
Brands that show up on YouTube are far more likely to earn citations, especially when your videos align with informational or instructional intent.
To capitalize on this:
- Create explainer videos that cover the fundamentals of your niche topics
- Produce how-to content with clear steps and voiceover guidance
- Include product demos or comparisons to capture commercial and review-based intents
- Use clean titles and descriptions that reflect natural language queries
- Add captions and timestamps to improve clarity and help the model parse structure
4. Track brand presence in AI answers
Traditional traffic metrics don’t tell the full story anymore.
Measuring your share of voice in AI answers is an essential reporting metric that helps you understand:
- Where your brand appears in AI responses
- Category winners
- Preferred content format
- Visibility changes across devices and markets
You can now track your AI performance with the AI Visibility tool in Moz Pro (now in open beta). It monitors brand mentions in generative search results, including GPT-4 responses, and shows how often your brand is cited.
How to get started
Go to Moz Pro, select AI Visibility on the left-hand menu, enter your brand name (e.g., Toyota), and click Check Visibility.
Add up to 5 related terms for your brand. For example, I’ve added Toyota Motors, Toyota cars, and Toyota vehicles as alternatives to the brand name, “Toyota.”
Next, input up to 3 competitors, along with their related terms (optional).
Add a list of topics for prompt generation and click Generate Prompts.
Moz AI generates prompts based on queries users are asking about your brand terms. Click Save & Continue to start tracking prompts.
This gives you a brand-level dashboard showing how often you’re cited, where you’re gaining ground, and which competitors are appearing more often. It’s a great way for SEOs to measure visibility beyond the blue links.
Moz’s AI Visibility is available on the standard, medium, and large pricing plans.
Concluding thoughts: Ranking isn’t the only factor that shapes visibility
AI Mode is changing how brands appear in search experiences. While ranking still plays a role, it’s no longer the only path to visibility. To maximize presence in AI responses, focus on building topic clusters that cover the breadth of a subject, expand your footprint across trusted domains, and structure content for clarity.