The signal

What changed.

Early data on AI search visibility is converging on a consistent finding: the brands showing up most in AI responses share the same characteristics. Quality content, mentions in authoritative publications, strong brand reputation, and genuine expertise.

The hypothesis

The emerging thesis we’re exploring.

We believe this is a silo problem above all else. A problem we're pretty familiar with. (It's the same fundamental problem that gave rise to growth as a function back in the early 2010s.)

Software companies presented a whole new surface area of problems that no existing team structure was set up to solve. The skills were mostly there; they just had to be reconfigured. AI search is following the same pattern.

New tactics and frameworks will emerge, but the core unlock is structural. The question isn't whether AI search is "really just SEO" or "actually a brand play" or something entirely new. That debate itself is the problem. It reinforces the very silos holding teams back. The answer is all of the above, repackaged into a unified discipline.

It's the thesis we built Saturation around. We think AI search is going to be the forcing function that finally brings brand and content into the growth fold. Disciplines that have been on their own islands for far too long.

The big picture

The structural shifts behind the hypothesis.

1. Each AI model behaves like a different era of search: Gemini pulls live results like Google, GPT relies on internal knowledge like Yahoo 2000s, Perplexity cites forward like Bing, Claude prioritizes depth over freshness. There is no single “AEO strategy.”​
2. AEO is a brand problem, not an SEO problem: LLMs replicate real-world popularity contests. PR and earned media build AI visibility more durably than content optimization.
3. Google rewrites your query into 5-40 sub-queries before answering: ranking for the original keyword only gives ~25% citation chance. Seven Google patents document pairwise LLM passage comparison behind the scenes.
4. AI chat queries average 23 words vs. 4.2-word keyword searches: users share full context rather than typing keywords. This isn’t SEO with longer queries. It’s a fundamentally different human behavior with different strategic implications.
5. Brand awareness has the second-strongest correlation with LLM visibility, behind domain rank. AI doesn’t discover brands: it picks up what the web is already paying attention to.​

What’s working

Tactics, experiments, and observations from the frontier.

• Track topics, not prompts: LLMs abstract queries internally, making prompt-level metrics noise. Topical authority is durable; prompt-specific position is not.
• External citations beat on-page authority for new brands: a test brand with zero DA hit 10% AI share of voice in 14 days using directories, Reddit, and LinkedIn alone.
• Real AI prompts look nothing like your keyword list: actual ChatGPT queries run 71% longer than marketers estimate and are 3x more problem-oriented. Real users don’t ask “best [category]”: they describe their situation, not a keyword.
• Third-party hosting silently transfers your citations: AI attributes content to where it discovers it, not who produced it.
• ChatGPT became Tally’s #1 referral source: the bootstrapped form builder says 25% of new signups now come from AI search discovery.​
• Up to 5x gap in AI Overview presence based on intent alone: how you frame the query matters more than the topic.​
• Information gain is mandatory: AI writes comprehensive content, so comprehensiveness is worth less. Segmented content drove 15.7x growth vs. decline for broad content.
• 82% of AI Overview citations come from deep interior pages: not homepages. Content in 50-150 word semantic units indexes better in vector embeddings.
• NPS predicts AI visibility better than domain authority: customers talking positively about you across Reddit, review sites, and forums is a more durable signal than backlinks.​

Frontier players

The people and teams building at the edge.

• ​Saturation: Purpose-built around the thesis that AI search is a silo problem, not a skills problem. Assembled strategists, AI specialists, brand builders, data scientists, and content architects under one roof.
• ​Kevin Indig: Ex-Shopify, G2, Atlassian. Takes a rigorously data-driven approach to AI search.
• ​Eli Schwartz: Author of Product-Led SEO. Ex-SurveyMonkey. Argues AEO isn’t a new discipline: what’s being called AEO is already covered by good brand and product strategy.
• ​Mike King: iPullRank founder. Patent-level analysis of AI Mode retrieval. Built Qforia.
• ​OtterlyAI: The GEO experiment lab. Public experiments with methodology and failures.
• ​Lily Ray: VP SEO, Amsive Digital. Tested which LLMs adopt fabricated claims.
• ​Ethan Smith: Graphite CEO. Enterprise-tier AEO/GEO pioneer.
• ​Britney Muller: Ex-Moz, ex-Hugging Face. Bridges ML concepts and SEO practice. Rare credibility when explaining why AI systems surface content the way they do.
• ​Kelsey Libert: Fractl co-founder. Original research on how brand mentions in third-party content drive AI citation.
• ​SearchPilot: GEO A/B testing methodology at scale.

The stack

Tools for building your AI growth engine.

AI search monitoring

• ​Peec AI: Brand citations across 10+ LLMs. $99/mo. Strong multi-platform coverage.
• ​OtterlyAI: Brand visibility + GEO auditing across 6 engines.
• ​PromptMonitor: Tracking + publisher outreach from cited sources.
• ​AthenaHQ: Ex-Google/DeepMind. YC W25. Shopify + GA integration.
• ​Profound: Enterprise. Sequoia-backed ($58.5M).
• ​Evertune: Monitoring + “AI Retargeting” on cited URLs. Ex-Trade Desk team.
• ​Knowatoa: “Search Console for AI.” Alerts when bots crawl your site.
• ​AmIOnAI: Actionable recs with personalized task plans.

Content & optimization

• ​Geoptie: All-in-one GEO toolkit. 7 free tools.
• ​Frase: Agentic SEO/GEO. 80+ AI skills.
• ​AI Rank Lab: AEO/GEO reports + bot tracking. Free score checker.
• ​LLMrefs: Tracker, 200+ AEO tools directory, and a free llms.txt generator.
• ​Rankability: Content and monitoring workflows for agencies.

Infra

• ​llms.txt: Open standard (robots.txt for AI). Free. Jeremy Howard (fast.ai).
• ​Kalicube Pro: Entity identity for Knowledge Graph + AI. 25B data points.
• ​Scrunch: See what AI bots see on your pages.
• ​Lily Ray’s Gemini RAG Tool: Free. See which sub-queries Gemini fires.
• ​SparkToro AI Prompt Intelligence: What your audience asks AI tools. Freemium.

Signal vs noise

The counter-arguments, and whether they hold up.

💨 “AEO is just SEO with a fancy name.”

🎯 Strong SEO foundations do predict AI citation. But 48% of AI citations now come from Reddit and YouTube, surfaces SEOs aren't optimizing for. The teams winning are treating it as a brand motion with a new-surface playbook, not choosing between them.

💨 “Quality content is overrated. Listicles were dominating AI citations.”

🎯 They were. Listicle citations in ChatGPT dropped 30% MoM in early 2026. Wikipedia nearly doubled. Reddit tripled. The listicle window was real, but it seems to be closing fast.

💨 “You don’t need a dedicated AI search strategy. Just do great marketing.”

🎯 The founder of Lorelight shut down his GEO platform after concluding the winners were all just "doing great marketing." He's not wrong. But it's not exactly a strategy, either. By that logic, why stop there? May as well just go back to telling people they don't even need great marketing, they should just build an amazing product and the customers will follow...

💨 “Small brands can’t compete in AI search. LLMs favor the incumbents.”

🎯 Domain rank has the strongest correlation with AI visibility (0.25). Sounds like incumbents win by default. But challenger brands like ShipStation are already outranking UPS in AI recommendations. And a test brand with zero domain authority hit 10% AI share of voice in 14 days. LLMs favor reputation, not size.

💨 “AI search is too early to invest in. Wait until it matures.”

🎯 ChatGPT has 800M+ weekly visitors. 25% of Tally’s signups come from AI discovery. It’s already a real acquisition channel. And if you understand how platform cycles work, the window to build presence is now.

Rabbit hole

If you wanna go even deeper.

• ​Eli Schwartz on the Intelligent Marketer Podcast: A case for why AEO is really just brand marketing.
• ​AI brand disambiguation: 75% of AI references cite the wrong company.
• ​GEO A/B testing: Will Critchlow on treating AI search as experimental science.
• ​The Next Great Distribution Shift: Brian Balfour on why ChatGPT is following the exact same platform cycle as Facebook, Google, and Apple.