Citation Differences in Google AI Overviews vs AI Mode

Google now runs two distinct AI search surfaces, and the data shows they behave like entirely separate systems when it comes to choosing which sources to cite. If your content strategy treats AI Overviews and AI Mode as interchangeable, you are leaving a significant portion of your potential AI visibility on the table.

This article breaks down what the research actually shows about how these two surfaces differ in citation behavior, why those differences exist, and what they mean for your content and optimization strategy.

Two Surfaces, One Search Engine

Google launched AI Overviews in May 2024 as an automatic feature that appears at the top of standard search results pages for informational queries. AI Mode followed in May 2025 as a separate, user-selected conversational search experience accessible via a dedicated tab.

Despite both being powered by Google’s Gemini model, they operate with different objectives. AI Overviews are designed to deliver fast, concise summaries for users who want a quick answer before deciding whether to click through. AI Mode is built for deeper exploration, supporting multi-turn conversations, follow-up questions, and longer-form synthesis across a wider range of sources.

The key question for SEOs and content marketers is whether optimizing for one automatically earns visibility in the other. The answer, backed by multiple large-scale studies, is a clear no.

The Citation Overlap Problem

The most striking finding in recent research is how rarely the two surfaces agree on which sources to cite.

Ahrefs analyzed 730,000 response pairs from September 2025 U.S. data and found that AI Mode and AI Overviews cited the same URLs only 13.7% of the time. When researchers narrowed the comparison to just the top three citations in each response, overlap rose slightly to 16.3%. That still means 87% of the time, these two systems pull from completely different sources to answer the same query.

SE Ranking’s independent study produced nearly identical results, finding only 10.7% URL overlap and 16% domain overlap between the two surfaces. Victorious ran a study across 1,540 queries and found that 77% of unique domains appeared in only one surface, with just 23% carrying across both. Not a single query in their dataset produced identical citation lists across both surfaces.

Google’s own documentation acknowledges that AI Mode and AI Overviews may use different models and techniques, which explains why their responses and linked sources vary even for the same query.

Why the Overlap Is So Low

The low citation overlap is not a bug. It reflects the fundamentally different retrieval objectives of each surface.

AI Overviews prioritize speed and precision. They are designed to surface a small, highly curated set of sources that can support a concise summary. The average AI Overview cites approximately 7.7 to 13.3 sources depending on the study, and those sources tend to skew toward well-known domains, video content, and core pages like homepages. YouTube is the most frequently cited source in AI Overviews, and they lean heavily on video and high-authority editorial content.

AI Mode prioritizes depth and breadth. It generates responses that are roughly four times longer than AI Overviews on average, and it cites a correspondingly wider range of sources. Ahrefs found that AI Mode averaged 3.3 entity mentions per response compared to just 1.3 for AI Overviews. AI Mode cited Wikipedia in 28.9% of responses versus 18.1% for AI Overviews, and cited Quora at 3.5 times the rate of AI Overviews. Health and medical sites were cited at roughly double the rate in AI Mode.

The two surfaces also differ in how reliably they attribute sources at all. Ahrefs found that 11% of AI Overviews contain no citations, compared to only 3% of AI Mode responses. AI Mode is more consistently attributive, which matters for brands trying to build citation presence.

Semantic Similarity vs. Source Divergence

One nuance that makes this situation particularly interesting is that the two surfaces often agree on what to say while disagreeing on where to source it.

Ahrefs found an average semantic similarity score of 86% between AI Mode and AI Overview responses for the same query, with nearly 90% of response pairs scoring above 0.8 on a scale where 1.0 indicates identical meaning. As Despina Gavoyannis, Senior SEO Specialist at Ahrefs, summarized: “Put simply, 9 out of 10 times, AI Mode and AI Overview agreed on what to say. They just said it differently and cited different sources.”

This means the two systems are reaching similar conclusions through different retrieval paths. They are not simply a short and long version of the same answer. They are independent systems that happen to converge on similar conclusions most of the time.

For content marketers, this distinction is critical. Being cited in one surface does not mean your content is the most authoritative source on a topic. It means your content matched the specific retrieval criteria of that surface at that moment.

Entity and Brand Mention Differences

The gap between the two surfaces is even more pronounced when it comes to brand and entity mentions.

Ahrefs found that 59.41% of AI Overviews contain no mentions of brands or entities, compared to 34.66% of AI Mode responses. This reflects the informational nature of most AI Overview queries, where named entities are not typically part of the answer.

AI Mode, by contrast, is far more brand-inclusive. Superlines research found that brands appear in approximately 90% of AI Mode responses, compared to only 43% of AI Overviews. For any brand trying to build awareness and consideration through AI search, AI Mode represents a significantly larger opportunity for brand-level visibility.

The entity relationship between the two surfaces also follows a consistent pattern. Ahrefs found that 61% of the time, AI Mode includes every entity that AI Overview mentioned, then adds more on top. This means that earning a brand mention in AI Overviews is a strong signal that you will also appear in AI Mode, but the reverse is not true. Many brands that appear in AI Mode responses are never mentioned in AI Overviews at all.

The Volatility Factor

Citation overlap is not just low between the two surfaces. It is also low within each surface over time.

Ahrefs found that 45.5% of AI Overview citations change when AI Overviews update, and AI Overview content changes approximately 70% of the time for the same query across repeated runs. SE Ranking found that AI Mode had overlapping results with itself just 9.2% of the time when the same query was tested three times. This means AI Mode is extraordinarily volatile, making it difficult to establish stable citation presence through any single piece of content.

This volatility has significant implications for measurement. A single snapshot of which sources are cited in AI Mode or AI Overviews tells you very little about your consistent visibility. Brands that want to understand their AI search presence need to track citation frequency across repeated tests, not just presence in a single run.

What This Means for Your Content Strategy

The research points to three strategic implications for teams optimizing for AI search visibility.

Treat AI Overviews and AI Mode as separate channels. Because citation overlap is so low, visibility in one surface does not predict or guarantee visibility in the other. Your monitoring, reporting, and optimization efforts need to account for both surfaces independently. Prioritize based on which surface your target audience is most likely to use for the queries that matter most to your business.

Match content format to surface behavior. AI Overviews favor concise, structured content that can be summarized quickly. Content that leads with a direct answer, uses clear headings, and provides structured data is more likely to be selected for the curated, short-form nature of AI Overviews. AI Mode favors longer, more exploratory content that covers a topic from multiple angles, includes entity-rich context, and supports the kind of multi-turn conversational queries that users bring to that surface.

Build citation authority across content types. The Ahrefs data shows that AI Overviews lean toward video content and core pages, while AI Mode draws more heavily from articles, forums, health sites, and Wikipedia. A citation strategy that relies solely on long-form blog content will optimize well for AI Mode but may underperform in AI Overviews. Brands that want presence across both surfaces need a content mix that includes video, structured entity pages, and editorial content.

The Content Placement Signal

One additional finding from Position Digital’s 100+ AI SEO statistics compilation is worth noting for content structure decisions. Research from Growth Memo found that 44.2% of all LLM citations come from the first 30% of text (the introduction), 31.1% come from the middle, and 24.7% come from the conclusion.

This pattern holds implications for both surfaces. Content that front-loads its most citable claims, data points, and direct answers is more likely to be extracted by either AI system. The BLUF (Bottom Line Up Front) principle that has long been recommended for AI optimization is supported by this citation distribution data.

Measuring Your Dual-Surface Visibility

Given the low overlap and high volatility across both surfaces, measurement requires a systematic approach.

For AI Overviews, Google Search Console provides some visibility into when your pages appear in AI-generated responses, though attribution is incomplete. Third-party tools including Semrush’s AI Visibility Index, Ahrefs’ Brand Radar, and SE Ranking’s AI Overview tracker offer more granular citation monitoring.

For AI Mode, measurement is more challenging because AI Mode sessions do not consistently generate referral traffic that can be attributed in standard analytics. Tools like Superlines, Profound, and Peec AI are designed specifically to track brand mentions and citation frequency across AI Mode and other AI search surfaces through systematic prompt testing.

The recommended approach is to build a prompt set of 30 to 100 queries representing your target topics, test them across both surfaces regularly, and track citation frequency as a core KPI alongside traditional traffic and ranking metrics.

The Strategic Takeaway

The data from multiple independent studies converges on a single conclusion: AI Overviews and AI Mode are not two versions of the same answer. They are distinct retrieval systems with different source preferences, different content type biases, different entity inclusion rates, and different volatility profiles.

For SEOs and content teams, this means the question is no longer just “how do I rank in AI search?” It is “which AI surface am I optimizing for, and does my content match what that surface is looking for?”

Brands that build monitoring and optimization systems for both surfaces now will identify citation patterns that competitors miss, adapt their content mix to match surface-specific preferences, and capture AI visibility opportunities that a single-surface strategy will consistently overlook.

The overlap between the two surfaces is 13.7%. The opportunity in the other 86.3% is where the competitive advantage lives.

A Practical Optimization Checklist for Both Surfaces

Given the distinct citation behaviors of each surface, the following framework helps content teams audit and optimize for both simultaneously.

For AI Overviews:

• Open every article with a 50-70 word direct answer to the primary query
• Use H2 and H3 headings that mirror common question formats
• Include structured data markup (FAQPage, HowTo, Article schema)
• Ensure the page is crawlable by Googlebot and indexed in Search Console
• Keep core pages (product pages, service pages, homepages) clean and entity-rich
• Embed or link to video content where relevant, as YouTube is the top-cited source in AI Overviews
• Target queries where Featured Snippets already appear, as 59.5% of AI Overviews co-occur with Featured Snippets

For AI Mode:

• Build long-form, exploratory content that covers a topic from multiple angles
• Include entity-rich context: named people, organizations, products, and their relationships
• Reference authoritative third-party sources including Wikipedia, academic studies, and industry reports
• Create content that supports multi-turn queries by anticipating follow-up questions
• Build topical clusters so AI Mode can draw from multiple related pages within your domain
• Earn mentions on forums, Q&A sites, and health/medical directories if relevant to your niche
• Optimize for question-based queries that invite deeper exploration rather than quick answers

For both surfaces:

• Front-load the most citable claims and data points in the first 30% of the article
• Maintain E-E-A-T signals: author credentials, publication dates, outbound citations to authoritative sources
• Keep content fresh with regular updates, as AI Overviews favor recent content
• Monitor citation frequency across both surfaces using dedicated AI visibility tools rather than relying on Google Search Console alone

The Competitive Opportunity

The low citation overlap between AI Overviews and AI Mode creates a competitive opportunity that most brands have not yet recognized. Most SEO and content teams are still optimizing for one surface, usually AI Overviews, because it is more visible and easier to track. AI Mode optimization remains underdeveloped across most industries.

Brands that build a dual-surface strategy now, while the field is still early, will establish citation patterns that become self-reinforcing over time. The Ahrefs finding that 61% of AI Mode responses include all entities from the corresponding AI Overview response suggests that earning AI Overview citations is a gateway to AI Mode visibility. But the reverse path, building AI Mode presence through entity-rich, exploratory content, can generate brand mentions in a surface where 90% of responses include brand names, versus only 43% for AI Overviews.

The data is clear: these are two different games with two different rules. The brands that learn both will own AI search visibility in a way that single-surface competitors simply cannot match.