[DON’T HAVE TIME? READ THIS]
Do you need a different SEO strategy for Google AI Mode?
No. Robby Stein (VP of Product for Google Search) confirmed AI Mode uses the same ranking signals as classic search, with five specific quality factors:
- Directly answers the user’s question – Intent match matters most
- High quality content – Built on 25 years of search quality systems
- Loads quickly – Site speed remains a ranking signal
- Original content – Not rehashed or duplicated information
- Cites sources – Attribution and credibility signals
Plus behavioral signals: if people click on it, value it, and come back to it, that content will rank in AI Mode.
How AI Mode works: The model generates dozens of queries under the hood, approximating what information people have found helpful for similar questions. It’s grounded in the same relevance, trust, and usefulness signals from classic search – not a separate system.
Bottom line: AI Mode is built on top of 25 years of search quality systems, not starting from scratch. The same SEO fundamentals that work for traditional search work for AI Mode.
Robby Stein sat down for an interview where he explained something most SEOs have been guessing about: how Google’s AI Mode actually determines which content to cite and recommend.
His answer was surprisingly specific. Five quality factors. Built on classic search signals.
AI Mode Is Built On 25 Years Of Search Quality Systems
The interviewer asked Stein how Google protects against AI hallucinations while maintaining search quality.
Stein’s answer cuts to the core of how AI Mode works: “While AI and generative AI in this way is frontier, thinking about quality systems for information is something that’s been happening for 20, 25 years. And so all of these AI systems are built on top of those”.
He explained that Google’s approach to determining what’s good information, which links are right, and what users would value is “all encoded in the model and how the model’s reasoning and using Google search as a tool to find you information”.
This is critical: AI Mode isn’t a separate system with different ranking criteria. It’s encoded with everything Google learned from classic search.
When someone searches for restaurants in a neighborhood, AI Mode leverages what people have been relying on Google for across years. “We kind of know what those resources are we can show you right there.”
The risk of hallucinations is managed by grounding AI answers in the same relevance, trust, and usefulness signals that have underpinned classic search for decades.
How Google Evaluates Helpfulness: Three Signal Categories
Stein explained that Google uses “a whole battery of things” to study helpfulness.
The evaluation happens across three categories:
Offline evaluation with real people – Human evaluators assess content quality outside of live search results.
Direct user feedback – Thumbs up/down ratings and whether people appreciate the information coming from AI Mode.
Behavioral signals – Are users using it more? Coming back? “Voting with their feet because it’s valuable”.
Stein noted something important: any one signal alone can lead you astray.
He specifically called out a trap that catches many products: high usage doesn’t always equal success. “In many products, if the product’s not working, you may also cause you to use it more.”
Google has a specific metric that tracks people trying to use it repeatedly for the same thing. They know that’s a bad signal – it means users can’t find what they need.
This carries over from classic search: success is judged by whether users are satisfied, not just whether they engage.
The Five Quality Factors For AI Mode Rankings
Stein was asked if SEO best practices still help for ranking in AI search.
His answer included five specific factors Google uses to determine if content meets quality and helpfulness standards:
1. Directly answering the user’s question – Content must solve the specific problem or need.
2. High quality – Based on 25 years of quality work around relevance, expertise, and trustworthiness.
3. Loads quickly – Site speed remains a ranking signal for AI Mode.
4. Original content – Not rehashed or duplicated from other sources.
5. Cites sources – Proper attribution and credibility signals.
Plus behavioral validation: “If people click on it, value it, and come back to it, that content will rank for a given question and it will rank in the AI world as well”.
Stein emphasized there’s “a very strong association to the quality work we’ve done over 25 years”.
The questions Google asks about content remain constant:
- Is this piece of content about this topic?
- Has someone found it helpful for the given question?
How Fan-Out Queries Work Under The Hood
Stein revealed a key mechanism for how AI Mode selects sources.
“The core mechanic is the model takes your question and reasons about it, tries to understand what you’re trying to get out of this. It then generates a fan-out of potentially dozens of queries that are being Googled under the hood”.
This fan-out approach is approximating what information people have found helpful for those questions.
The result: AI Mode can surface broader diversity of content than traditional search “because it’s doing research for you under the hood”.
If your content aligns well across these fan-out queries – addressing the main question plus related sub-topics – it becomes more eligible for AI-generated answers.
This reinforces the importance of comprehensive topic coverage rather than narrow keyword targeting. AI Mode cross-checks information using multiple queries to ensure trusted sources align.
What This Means For SEO Strategy
Stein’s explanation confirms what Google has been saying: the fundamentals don’t change.
AI Mode relies on classic search signals. The five factors he named aren’t new – they’re established quality principles applied to a new interface.
The shift is in how content gets evaluated. Instead of matching a single query, your content might get pulled into answers for dozens of related queries generated during the fan-out process.
This favors:
- Comprehensive coverage – Addressing questions from multiple angles
- Semantic relevance – Related concepts, not just exact keywords
- Topic authority – Consistent quality across related subjects
The old approach of optimizing individual pages for specific keywords becomes less relevant. AI Mode looks at whether your content satisfies the underlying information need across multiple query variations.
Why Historical Performance Matters More
One factor Stein emphasized: Google knows what resources people have been relying on “for all these years”.
This means historical performance, domain trust, and content consistency matter more than short-term SEO tricks.
If your site has a track record of providing helpful information on a topic, that history is encoded into the model’s understanding of which sources to trust.
New sites or sudden content pivots face an uphill battle. The model leverages decades of user behavior data to identify reliable sources.
This aligns with Google’s broader E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness), but with behavioral validation built in.
The Continuity Message
Stein’s most important point: AI Mode represents continuity, not disruption.
“All of these AI systems are built on top of” 25 years of search quality work.
The same content strategies that worked for traditional search work for AI Mode:
- Answer user questions directly
- Create high-quality, original content
- Ensure fast load times
- Cite credible sources
- Earn clicks and engagement
The interface changed. The ranking fundamentals didn’t.
SEO professionals focusing on real user value, authoritative insights, and content depth will succeed in AI Mode the same way they succeed in classic search.
Stein’s five factors aren’t a new playbook. They’re confirmation that the existing playbook still works.
