Google search is no longer a single AI system with a single rulebook. A study by OtterlyAI reveals that Google is quietly running two parallel AI search experiences: AI Overviews, which are the familiar AI summaries at the top of search results, and AI Mode, the new AI-powered chat-like search interface.
In this study, we ran 100 of Germany’s top Googled search queries through both AI systems and tracked more than 30,000 citations. The patterns are clear:
- Google AI Mode behaves like an always-on research assistant that pulls from a wide range of sources and heavily features brand-owned content.
- Google AI Overviews behaves more like a curated answer layer that appears less often, leans into YouTube, Wikipedia, official sites, and often prefers third-party coverage over a brand’s own pages.
The results reveal two fundamentally different approaches to sourcing information, with major implications for how brands should approach their AI search strategy.
📌 Key Takeaways
- 49% vs 100% Trigger Rate: AI Overviews only appeared for 49 of 100 queries we ran. However, AI Mode triggered 100 times for 100 search queries. This demonstrates how AI Overviews work, as they are not triggered for every search query. More importantly, it tells us that AI Mode provides double the query coverage, which is critical for brands tracking visibility.
- 6× more citations when both trigger: or queries where both systems respond, AI Mode averages 310 citations vs 51 for AI Overviews. From these results, it’s clear that AI Mode performs deeper, multi-source research.
- Source Diversity: AI Mode draws from 3,621 unique domains vs 615 for AI Overviews. However, concentration metrics show AI Overviews is actually MORE concentrated (21.8% in top 10 domains vs 18.8%).
- Source Type Preferences: Retail sites represent 14.2% of AI Mode citations vs only 4.4% in AI Overviews. Brand-specific queries drive massive citation clusters. AI Mode heavily cites Google properties (8.8% vs 3.6%), including support.google.com, play.google.com, and translate.google.com.
The Headline Numbers: AI Overviews triggered for 49% of all search queries
AI Mode generated 30,353 citations across our test queries. AI Overviews produced 2,498. That’s a 12× difference in raw volume, but the real story is more nuanced.
However, AI Mode triggered on 100 of our 100 test queries. AI Overviews only appeared for 49. Based on those results,AI Mode is essentially always on, while AI Overviews remains selective about when it shows up. Put simply it’s up to Google to decide if an AI Overview is triggered or not.
When both systems do respond to a query, AI Mode averages 310 citations versus 51 for AI Overviews. AI Mode also draws from a much larger pool of sources, citing 3,621 unique domains compared to 615 for AI Overviews.
Different Systems, Different Winners
The most striking finding, though, is the different ways that these systems treat the same domains. We calculated each domain’s “share” of total citations and compared them across both systems.
The share difference shows which system gives each domain more relative visibility.
- Positive = favored by AI Overviews.
- Negative = favored by AI Mode.
- Youtube is the way to win in AI Overviews
YouTube captures 6.33% of all AI Overviews citations but only 2.02% in AI Mode. That’s a 3× advantage in AI Overviews. Video content clearly matters more when Google is being selective about sources. - AI Mode is where Google prefers Google
Google.com shows the opposite pattern: 3.41% share in AI Mode versus 0.04% in AI Overviews. AI Mode leans heavily on Google’s own properties, including Play Store, Translate, and support documentation. AI Overviews barely touches them. - Wikipedia is your universal source of truth
Wikipedia performs well in both systems, capturing 3.63% of AI Mode citations and 4.76% of AI Overviews. It’s the one consistent winner regardless of which AI experience triggers. - For sensitive and regulated topics, borrow authority from official sources.
Government and official sources dramatically over-index in AI Overviews. The German federal cybersecurity agency (bsi.bund.de) captures 1.04% of AI Overviews citations versus just 0.01% in AI Mode. That’s a 100× difference in relative visibility. - B2C wins in AI Mode, not in AI Overviews
B2C and e-commerce sites show the reverse pattern. Brands like DHL, Rossmann, and Deutsche Bahn appear heavily in AI Mode but are nearly invisible in AI Overviews. Retail sites overall capture 14.2% of AI Mode citations versus 4.4% in AI Overviews.
Brand Queries Tell a Clear Story
We looked specifically at queries containing brand names to see how often each system cites the brand’s own website.
For DHL queries, 58.7% of AI Mode citations went to dhl.de. In AI Overviews, that dropped to 15.4%. Rossmann saw 62.9% brand attribution in AI Mode versus 25% in AI Overviews. MediaMarkt and Spiegel received zero citations to their own domains in AI Overviews while capturing over 55% in AI Mode.
The pattern is consistent: AI Mode favors brand-owned content for brand queries. AI Overviews prefers third-party sources, even when someone is explicitly searching for a specific brand.
This has direct implications for brand control. In AI Mode, comprehensive owned content translates to visibility. In AI Overviews, third-party mentions and authoritative citations carry more weight.
What This Means for Your AI Search Strategy
The divergence between these systems creates a new optimization challenge. A single approach won’t work for both.
Google AI Mode:
For e-commerce and B2C brands, AI Mode is the priority. Your product pages, category content, and support documentation all get pulled into AI Mode responses. The system rewards depth and comprehensiveness. Build out your site architecture, add detailed product information, and don’t neglect FAQ content. Brand queries in AI Mode heavily favor your own domain.
For local and service businesses, AI Mode’s 100% trigger rate means more opportunities to appear. Focus on comprehensive local content and review management. AI Mode pulls from a wider range of sources including forums and UGC platforms that don’t appear in AI Overviews at all.
Google AI Overviews:
For publishers and media companies, video content is non-negotiable for AI Overviews visibility. The study shows that brands that have YouTube in their content strategy, get cited 3x more in AI Overviews than brands that don’t. This suggests that Google is prioritizing multimedia content formats. Consider creating video versions of your most important articles, especially for how-to and explainer content as informational content has been hit the worst by the recent surge of zero-clicks. Written web-content still performs well in AI Mode, so you’re hedging across both systems.
For B2B and SaaS companies, AI Overviews rewards authority signals. Technical documentation, definitive “what is” content, and partnerships with government or industry bodies all over-index in AI Overviews. Adobe, Okta, and Microsoft support pages all capture disproportionate share in AI Overviews compared to AI Mode. Build glossaries, educational hubs, and reference content that establishes your expertise.
The Bigger Picture
These findings point to two distinct philosophies at work.
AI Mode seems to operate like an agentic research assistant. It casts a wider net, pulls from hundreds of sources per query, and seems to prioritize comprehensiveness over curation. It favors brand-owned content, Google’s own properties, and deep site architectures.
AI Overviews operates like a curated answer engine. It’s selective about when it appears, draws from fewer but more authoritative sources, weights video content, government sites, and third-party validation more heavily. It deprioritizes brand-owned content even for brand queries.
Neither approach is objectively better. They serve different user intents. AI Mode seems designed for research and exploration. AI Overviews seems designed for quick, trusted answers.
What You Should Do Now
1. Start tracking both systems separately. The 49% trigger rate gap means AI Overviews data alone gives you an incomplete picture of your AI search visibility. If a query triggers AI Mode but not AI Overviews, you need to know whether you’re being cited. Tools like OtterlyAI can monitor your visibility across both experiences and alert you to gaps.
2. Audit your Wikipedia presence immediately. It’s one of the few sources that performs consistently well in both systems, capturing 3.6% to 4.8% share regardless of which AI experience triggers. Errors, outdated information, or gaps in your Wikipedia coverage will propagate across all AI search experiences. This is foundational work that pays dividends everywhere.
3. Create video content for your highest-value topics. YouTube’s dominance in AI Overviews makes video a competitive moat, especially for informational and how-to queries. If you’re only publishing text content, you’re leaving AI Overviews visibility on the table. Start with your top-performing articles and create video companions.
4. Build comprehensive owned content across your entire site. AI Mode rewards depth and breadth. More pages covering more topics with more detail and more structured data means more opportunities to be cited when AI Mode conducts its research. Don’t just optimize your homepage; build out category pages, product pages, support documentation, and FAQ sections.
5. Earn authoritative third-party mentions. AI Overviews weights external validation heavily. PR coverage in reputable publications, industry recognition and awards, and citations from government or educational sources all compound your visibility in the curated AI Overviews experience. This is slower work but creates lasting competitive advantage.
6. The era of optimizing for a single search experience is ending. Google now runs multiple AI systems with different source preferences, different trigger conditions, and different implications for brand visibility. The winners will be organizations that understand both systems and build content strategies that perform across the full spectrum of AI search.
This analysis is based on 100 popular German search queries tracked through OtterlyAI in October-November 2025, generating 30,353 AI Mode citations and 2,498 AI Overviews citations across 3,621 unique domains.