How AI media partnerships influence your brand visibility in genAI: Research
In a recent study, Search Engine Land and Fractl found that 82% of consumers find AI-powered search more helpful than traditional SERPs.
While the emergence of generative engine optimization (GEO) has marketers in a frenzy to own the latest industry keyword, agency thought leaders are finding common ground.
Whether it’s Google or generative AI, brand visibility still comes down to two things:
Together, these build a digital footprint of authority that both algorithms and knowledge graphs use to surface your brand.
Here’s the rub: most brands are still playing a one-dimensional game.
They’re optimizing content hubs around FAQs and target market needs while ignoring offsite authority signals like brand mentions, which strengthen visibility across genAI search.
To highlight the need for a two-prong approach, my Fractl co-founder, Dan Tynski, scraped 8,090 keywords across 25 verticals to compare citations between Google’s AI Overviews and LLMs (GPT, Claude, Gemini, etc.) for identical queries.
The result should give every marketer pause: only 7.2% of domains appeared in both systems.
Of the 22,410 unique domains we identified:
So, what does this mean for marketers?
Unsurprisingly, Google’s AI system heavily favors the same reputable domains we’re used to dominating the SERPs.
High-authority websites with robust content portfolios and strong backlink profiles dominated Google’s AI Overviews.
These 15,848 domains were largely represented by:
The 4,951 LLM-exclusive domains tell a different story.
They’re significantly smaller – three times fewer than AI Overviews – and reflect what LLMs actually value:
Ultimately, foundation models seem to prioritize publishers that provide topic depth over topic breadth, and educational value and conceptual clarity over traditional web authority signals.
Your DA 90 site might be invisible to ChatGPT if it doesn’t clearly and effectively explain concepts, rather than just ranking well with authority.
Based on our analysis, here are the five laws that determine whether your content gets cited across AI platforms:
Strong editorial standards and human fact-checking (think NPR, NYT) get linked, cited, and re-crawled more often.
Repeated crawls mean their phrasing becomes the “default” language models lean on when answering.
Action items
Models love patterns, so think step-by-step how-tos, definition blocks, ordered lists, and comparison tables that use standard layouts that are machine-friendly.
Action items
Niche experts (Edmunds for cars, Mayo for health, etc.) flood the web with highly structured, topic-dense updates.
Models learn, “When it’s cars, go to Edmunds; when it’s investing, go to Investopedia.”
Action items
Big outlets get quoted, reprinted, aggregated – each copy reinforces word patterns and narrative frames.
One AP pickup becomes 200 local clones. Now that wording is everywhere in the model’s diet.
Syndication is how one fact becomes “the fact” in AI.
Action items
Training corpora over-index on U.S. English, ad-supported news, and commercial publishers due to current partnerships.
Non-U.S., non-English, academic, or NGO sources are underweighted – leading to culturally narrow answers.
Action item
Get the newsletter search marketers rely on.
See terms.
Beyond the content strategies that shape the knowledge graphs of generative AI platforms, I also wanted to understand which media conglomerates are being cited most frequently.
Since I oversee our agency’s earned media team, it felt imperative to prioritize my digital PR team’s targeting of the most cited publishers within each client’s niche.
I set off to research the “AI media partnerships” and licensing agreements that OpenAI, Perplexity, and others had arranged over the last few years.
These partnerships pull three primary levers that shape a model’s internal neural network and knowledge graph about your industry and brand:
When a publisher network becomes a model partner, it stops being just another website the AI can read – it becomes a trusted source the model actively learns from and reuses.
Over time, that content becomes a landmark inside the system’s knowledge map, shaping how the model understands topics, brands, and credibility.
As AI assistants build more structured “answer pipelines,” these publisher networks hold a real advantage – their stories are more likely to be cited, repeated, and remembered.
As a brand, if your story lives outside these publisher networks, you’ll spend more time and budget fighting to be represented as an authority in generative answers.
Dig deeper: Tracking AI search citations: Who’s winning across 11 industries
As generative AI continues to train on public web data, not all platforms are created equal.
If trust becomes the new currency of the internet, then the source of that data, not just its scale, determines its long-term value in model training.
Reddit, Quora, and other UGC-heavy platforms may dominate today’s AI citations.
But they’re also the most vulnerable to contamination, bias loops, and synthetic noise.
As the signal-to-slop ratio worsens, these sources could face a credibility correction once models start weighting for provenance, diversity, and verifiability.
We built a forecasting trusted sources framework to quantify which publisher types are best positioned to retain influence in genAI ecosystems.
By scoring platforms across seven trust signals – from scarcity and verifiability to legal clarity and longevity – we can forecast which media environments are most likely to feed the next generation of training data.
In short: brands that invest in credible, human-authored, legally clear coverage today will become the foundational voices tomorrow’s models rely on.
The brands that dominate generative engine optimization in 2026 won’t chase rankings. They’ll architect authority.
They’ll prioritize strategies that build cross-channel visibility into a digital footprint strong enough to influence knowledge graphs, algorithms, and audiences alike.
If you want to be that brand, here’s your action plan:
Generative visibility rewards depth, not breadth.
The top performers – WebMD, All Recipes, U.S. News – own their verticals completely.
Identify your subcategory and build the most complete knowledge base in it.
LLMs cite what they can parse.
Standardize your content templates – definitions, FAQs, how-tos, comparison tables – and use clear schema markup.
What’s good for machines tends to be good for humans, too.
Engineer syndication paths for every data study.
Every syndicated mention, embedded chart, and reused quote compounds authority.
Make it easy for journalists and genAI platforms to reuse your language, charts, and insights.
One pickup can become 200 citations across the web.
AI models favor content from trusted media groups (TIME, FT Group, Guardian Media, Axel Springer).
Earn placements within those networks to increase your odds of inclusion in retrieval pipelines.
U.S. outlets dominate AI training data, but multilingual and regional content can fill cultural gaps.
Localize and translate your assets to improve visibility in global models.
Peer-reviewed data, transparent sourcing, and expert-authored content outperform SEO-optimized filler in training value.
In genAI search, credibility is the new ranking factor.
The next evolution of search isn’t a race for keywords.
It’s a race for context, credibility, and coverage.
Build your digital footprint of authority through brand mentions now, while your competitors are still optimizing the 101 of site architecture and content hubs.
Once they understand the new rules of discoverability, the knowledge graphs will already know who to trust.
Continue reading...
In a recent study, Search Engine Land and Fractl found that 82% of consumers find AI-powered search more helpful than traditional SERPs.
While the emergence of generative engine optimization (GEO) has marketers in a frenzy to own the latest industry keyword, agency thought leaders are finding common ground.
Whether it’s Google or generative AI, brand visibility still comes down to two things:
- The depth of your original subject matter expertise.
- The breadth of your brand mentions.
Together, these build a digital footprint of authority that both algorithms and knowledge graphs use to surface your brand.
Here’s the rub: most brands are still playing a one-dimensional game.
They’re optimizing content hubs around FAQs and target market needs while ignoring offsite authority signals like brand mentions, which strengthen visibility across genAI search.
To highlight the need for a two-prong approach, my Fractl co-founder, Dan Tynski, scraped 8,090 keywords across 25 verticals to compare citations between Google’s AI Overviews and LLMs (GPT, Claude, Gemini, etc.) for identical queries.
The result should give every marketer pause: only 7.2% of domains appeared in both systems.
Of the 22,410 unique domains we identified:
- 15,848 domains (70.7%) appeared exclusively in Google AI Overviews.
- 4,951 domains (22.1%) appeared exclusively in LLM foundation models.
- 1,611 domains (7.2%) appeared in both systems.
So, what does this mean for marketers?
1. Google’s AI Overviews still favor the old guard
Unsurprisingly, Google’s AI system heavily favors the same reputable domains we’re used to dominating the SERPs.
High-authority websites with robust content portfolios and strong backlink profiles dominated Google’s AI Overviews.
These 15,848 domains were largely represented by:
- Established news and information sites (BBC, Yahoo, CNN).
- Educationally leaning social sites (Reddit, YouTube).
- Authoritative reference sources (Wikipedia, peer-reviewed journals).
- Government and institutional websites (.gov, .edu).
2. Publishers driving generative AI brand visibility and authority
The 4,951 LLM-exclusive domains tell a different story.
They’re significantly smaller – three times fewer than AI Overviews – and reflect what LLMs actually value:
- Investigative journalism from mainstream news publishers covering timely news pulled by RAG searches (e.g., USA Today, CNBC, The New York Times).
- Niche, vertical experts that demonstrate deep subject matter expertise within a specific vertical (e.g., Edmunds, Investopedia, All Recipes, Wired).
- Educational platforms and communities optimized for learning (Reddit, Github, Coursera, Khan Academy, University hubs).
- Authoritative industry data portals, such as peer-reviewed journals, patents, and standards, court and government transcripts.
Ultimately, foundation models seem to prioritize publishers that provide topic depth over topic breadth, and educational value and conceptual clarity over traditional web authority signals.
Your DA 90 site might be invisible to ChatGPT if it doesn’t clearly and effectively explain concepts, rather than just ranking well with authority.
Based on our analysis, here are the five laws that determine whether your content gets cited across AI platforms:
Authority creates trust loops in LLMs
Strong editorial standards and human fact-checking (think NPR, NYT) get linked, cited, and re-crawled more often.
Repeated crawls mean their phrasing becomes the “default” language models lean on when answering.
Action items
- Publish sourced, well-edited, standards-based content.
- Earn links from high-authority newsrooms.
- Maintain freshness to trigger re-crawls.
If it’s easy to skim, it’s easy to train
Models love patterns, so think step-by-step how-tos, definition blocks, ordered lists, and comparison tables that use standard layouts that are machine-friendly.
Action items
- Standardize your article templates.
- Mark up with headings, schema, bullets, FAQs, and TL;DR summaries.
Vertical specialists train the model’s mental map
Niche experts (Edmunds for cars, Mayo for health, etc.) flood the web with highly structured, topic-dense updates.
Models learn, “When it’s cars, go to Edmunds; when it’s investing, go to Investopedia.”
Action items
- Build the definitive source of knowledge for your industry.
- Create step-by-step how-tos, definition blocks, comparison tables – the structured content that models love to cite.
- Pitch experts to the media to reinforce association and expertise.
Repetition and syndication = Statistical gravity
Big outlets get quoted, reprinted, aggregated – each copy reinforces word patterns and narrative frames.
One AP pickup becomes 200 local clones. Now that wording is everywhere in the model’s diet.
Syndication is how one fact becomes “the fact” in AI.
Action items
- Target syndication networks, wire partners, and data stories that invite re-use.
- Provide copy-and-paste-friendly assets (charts, stat bullets, embed codes).
U.S. commercial bias skews the knowledge lens
Training corpora over-index on U.S. English, ad-supported news, and commercial publishers due to current partnerships.
Non-U.S., non-English, academic, or NGO sources are underweighted – leading to culturally narrow answers.
Action item
- Earn media mentions in the same publishers generative AI platforms are using to train their models.
Get the newsletter search marketers rely on.
See terms.
3. Understanding how AI media partnerships influence brand visibility
Beyond the content strategies that shape the knowledge graphs of generative AI platforms, I also wanted to understand which media conglomerates are being cited most frequently.
Since I oversee our agency’s earned media team, it felt imperative to prioritize my digital PR team’s targeting of the most cited publishers within each client’s niche.
I set off to research the “AI media partnerships” and licensing agreements that OpenAI, Perplexity, and others had arranged over the last few years.
These partnerships pull three primary levers that shape a model’s internal neural network and knowledge graph about your industry and brand:
- Coverage: The legality, depth, and recency of the archives it can crawl and reuse.
- Context: The frequency with which those sources appear across pretraining data, retrieval indexes, evaluation sets, and safety workflows.
- Credibility: The confidence weight a system assigns when cross-checking and ranking those sources at answer time.
When a publisher network becomes a model partner, it stops being just another website the AI can read – it becomes a trusted source the model actively learns from and reuses.
Over time, that content becomes a landmark inside the system’s knowledge map, shaping how the model understands topics, brands, and credibility.
As AI assistants build more structured “answer pipelines,” these publisher networks hold a real advantage – their stories are more likely to be cited, repeated, and remembered.
As a brand, if your story lives outside these publisher networks, you’ll spend more time and budget fighting to be represented as an authority in generative answers.
Dig deeper: Tracking AI search citations: Who’s winning across 11 industries
4. Getting off the Reddit hype train
As generative AI continues to train on public web data, not all platforms are created equal.
If trust becomes the new currency of the internet, then the source of that data, not just its scale, determines its long-term value in model training.
Reddit, Quora, and other UGC-heavy platforms may dominate today’s AI citations.
But they’re also the most vulnerable to contamination, bias loops, and synthetic noise.
As the signal-to-slop ratio worsens, these sources could face a credibility correction once models start weighting for provenance, diversity, and verifiability.
We built a forecasting trusted sources framework to quantify which publisher types are best positioned to retain influence in genAI ecosystems.
By scoring platforms across seven trust signals – from scarcity and verifiability to legal clarity and longevity – we can forecast which media environments are most likely to feed the next generation of training data.
In short: brands that invest in credible, human-authored, legally clear coverage today will become the foundational voices tomorrow’s models rely on.
5. What GEO means for your 2026 SEO strategy
The brands that dominate generative engine optimization in 2026 won’t chase rankings. They’ll architect authority.
They’ll prioritize strategies that build cross-channel visibility into a digital footprint strong enough to influence knowledge graphs, algorithms, and audiences alike.
If you want to be that brand, here’s your action plan:
Choose your lane and own it
Generative visibility rewards depth, not breadth.
The top performers – WebMD, All Recipes, U.S. News – own their verticals completely.
Identify your subcategory and build the most complete knowledge base in it.
Engineer for structure
LLMs cite what they can parse.
Standardize your content templates – definitions, FAQs, how-tos, comparison tables – and use clear schema markup.
What’s good for machines tends to be good for humans, too.
Leverage ‘statistical gravity’
Engineer syndication paths for every data study.
Every syndicated mention, embedded chart, and reused quote compounds authority.
Make it easy for journalists and genAI platforms to reuse your language, charts, and insights.
One pickup can become 200 citations across the web.
Monitor your partnerships ecosystem
AI models favor content from trusted media groups (TIME, FT Group, Guardian Media, Axel Springer).
Earn placements within those networks to increase your odds of inclusion in retrieval pipelines.
Think global, not just Google
U.S. outlets dominate AI training data, but multilingual and regional content can fill cultural gaps.
Localize and translate your assets to improve visibility in global models.
Rebuild for trust
Peer-reviewed data, transparent sourcing, and expert-authored content outperform SEO-optimized filler in training value.
In genAI search, credibility is the new ranking factor.
Authority wins the race
The next evolution of search isn’t a race for keywords.
It’s a race for context, credibility, and coverage.
Build your digital footprint of authority through brand mentions now, while your competitors are still optimizing the 101 of site architecture and content hubs.
Once they understand the new rules of discoverability, the knowledge graphs will already know who to trust.
Continue reading...