We didn’t set out to find a keyword list. We set out to understand what actually happens in the moment before someone buys: the exact question they type, or speak, or paste into a chatbot, and what determines whether your brand gets named in the answer.
What we found was a pattern about trust. The brands that surface in AI search aren’t the ones with the most pages or the biggest ad budgets. They’re the ones with the most credible, human conversation happening about them across the open web. And influencer content has quietly become one of the strongest sources of that conversation. Not just as an awareness play, but as infrastructure for AI discovery.
We proved this with Self Financial, a credit-building fintech. The system we built drove a +31% lift in share of voice and a +155% lift in branded search in a single quarter. We’re now running the same playbook for a family and baby wellness brand. Here’s what the research surfaced, and exactly how the system works.
Key Takeaways
- Trust is the ranking signal in AI search. Answer engines like ChatGPT, Perplexity, and Google’s AI Overviews build their picture of a brand from credible third-party conversation, not just your website. Answer Engine Optimization (AEO) is the practice of earning a place inside that answer.
- Reddit is structurally wired into AI. It accounts for 40.11% of all LLM citations (per Semrush) and is formally connected to both Google and OpenAI through data-sharing agreements worth a combined $130M. Community conversation is what AI is trained to trust.
- Influencer content is an AEO source, not just an awareness play. A creator’s caption is now a citation candidate. The same content that earns trust on social is the content AI learns from and quotes.
- We have the receipts. Self Financial saw a +31% share of voice QoQ, +155% branded search QoQ, +41% YoY Reddit chatter, 3x iROAS, and 86% over performance target.
- It’s a repeatable system. Listen to how customers talk, map that language into creator briefs and your own pages, then measure AI share of voice. The answer to what to optimize toward lives in the conversation, not a keyword list.
- An enterprise agency partner should reduce your team’s load, not add to it. The right partner handles strategy, creator management, content review, compliance, and reporting, so your internal team focuses on high-level direction, not logistics.
What Were Brands Actually Searching For?
The clearest illustration is a single caption. A typical influencer post sounded like brand copy: “So excited to use this app!!” That says nothing a person would search for and gives AI nothing to work with. The version that earned visibility sounded like a person answering a real question: “I used it to build credit with small monthly payments. Here’s how it works.” Same product. Same creator. The first is a scroll-past. The second is something an answer engine can lift and quote.
That gap matters more every quarter. About 36% of consumers (and 47% of Gen Z) say they have discovered a new product or brand through ChatGPT (Adobe, 2025), and an additional 41% of Gen Z now turn to social platforms as their primary search tool, per our own market intelligence research. They rarely type brand names. They type problems: “just graduated, how do I build credit?” or “my baby has colic at 2am and nothing is working.” The brand that gets named is the one whose conversation already matches the question.
Why Reddit and Creator Content Are the Signals AI Trusts
Because AI engines don’t browse feeds. They read authority signals across the open web and decide whether to put your name in the answer. AEO is the practice of earning that placement. Its close cousin, GEO (Generative Engine Optimization), applies the same logic specifically to generative tools like ChatGPT and Perplexity. Both reward the same things: relevance to the real question, clean content structure AI can parse, fresh and frequently updated material, and credible third-party sources.
Reddit has been the single most-cited web domain across large language models, accounting for 40.11% of LLM citations in June 2025 (per Semrush). That’s not a coincidence. Google signed a $60M data-sharing agreement with Reddit, and OpenAI followed with a $70M annual licensing deal, formally wiring Reddit conversations into the core of both AI training and real-time search visibility. Community consensus and lived experience are exactly what these models are designed to surface.
That visibility isn’t permanent, though. In one Semrush study of more than 100 million AI citations, Reddit’s share swung from roughly 60% to around 10% in a matter of weeks before recovering. AI search visibility is not a trophy you win once. As we’ve argued about influencer marketing performance more broadly, the loop has to compound continuously, or it stops working.
This is where influencer content fits: right in the middle of it. When a creator explains what your product does in their own words, in response to a real question, they’re doing two jobs at once. Reaching an audience and feeding the corpus AI engines learn from. A creator video that never sends a single click to your site can still teach ChatGPT what you do, who you’re for, and why you’re worth recommending. And as discovery has shifted from intent-based search to interest-graph-driven surfaces, that content reaches people long before they’ve formed a query at all.
How to Get Found in AI Search
You build a loop, not a launch.
Listen. Use social listening to learn the real language of your category: the phrases customers use when no brand is in the room. Reddit, X, and YouTube at minimum.
Map intent. Reverse-engineer those phrases into your SEO, GEO, and AEO strategy. Find the white space (the questions your competitors aren’t answering) and go own that ground first.
Brief creators. Have creators talk about the product the way it’s organically discussed. Problem-first. Specific. Answerable. Authenticity is the format AI rewards because it’s the format people trust.
Optimize your own pages and retail listings. Answer real questions on your site with clear structure and schema. Don’t overlook Amazon and Walmart: both rank in the top-cited LLM domains, which means product listing copy is also AEO territory.
Measure AI share of voice. Track how often you’re cited in ChatGPT, Perplexity, and AI Overviews for the questions your customers actually ask. Refresh what’s working. Retire what isn’t.
Plenty of tools can tell you where you rank. Far fewer can tell you what to optimize toward, because that answer lives in the conversation. Not in a generic keyword list.
How Self Financial Grew It’s AI Search Visibility
Before a single creator was briefed, The Shelf delivered Self Financial a market intelligence report built from our proprietary social listening methodology. It identified the key talking points their consumer base actually cared about, language the brand wasn’t using but their customers already were. That intelligence became the strategic foundation for everything that followed.
The insight was deceptively simple. Customers discussing credit-building weren’t using brand language. They were describing real problems: “I just moved to the U.S. with no credit history,” “I need to prove I can manage money,” “I want to build credit without a credit card.” Those aren’t keywords. They’re the shape of the conversation AI is trained to recognize and cite.
The work was built on one principle: credible third-party conversation is the only ranking signal that matters in AI search. Not your website. Not your ad budget. These are trusted human conversations happening across the open web, on Reddit, in creator captions, in customer reviews. They are what answer engines learn from and quote.
So the strategy flipped. Instead of starting with what Self wanted to say, it started with what people were actually asking. Creators weren’t briefed with product benefits. They were briefed with real customer questions and given permission to answer them the way they actually would, problem-first, plain language, no hype. The same authenticity that builds trust on social media is exactly what AI engines are built to surface.
Because here’s the thing: a creator video that never sends a single click to Self’s site can still teach ChatGPT what the product does, who it’s for, and why it matters. And as discovery has shifted away from branded search toward AI-powered answers, that’s often where the real value lives.
The Results
+31% growth in AI share of voice, quarter over quarter +155% lift in branded Google search, quarter over quarter. +41% YoY increase in Reddit mentions (the most-cited domain across LLMs) 3x return on ad spend, 86% above performance target.
The mechanism is a loop, not a launch: conversation drives AI citations. Citations drive curiosity. Curiosity drives searches and branded discovery. And that loop compounds each quarter, or it stops working if you stop feeding it.
Frequently Asked Questions (FAQ’s)
What is AEO (Answer Engine Optimization)? AEO is the practice of structuring and earning content so that AI-powered answer engines (ChatGPT, Perplexity, Google’s AI Overviews) select your brand as part of their generated answer. Where traditional SEO competed for a ranking position, AEO competes for inclusion inside the answer itself.
How does influencer content help with AI search visibility? Creator content answers real questions in human language across trusted platforms, which is exactly what AI engines learn from and cite. The key is briefing creators to talk the way customers actually search, not in brand hype language. The same content that earns trust on social is the content AI lifts as an answer.
How do you measure AI search visibility? The core metric is AI share of voice: how often your brand is cited when customers ask AI engines questions in your category. Track it by running those questions inside ChatGPT, Perplexity, and AI Overviews regularly, noting whether you’re named and which sources are cited. We complement this with branded Google search lift and Reddit chatter volume, which move together in a healthy system.
The Shelf’s View
The brands that will win AI search are the ones treating influencer content as more than a media buy. They’re treating it as a trust-building system: one that generates the kind of human, contextual, question-answering content that AI engines were built to surface. They’re running social listening before creative, not after. They’re briefing creators with the language customers actually use. And they’re measuring toward AI share of voice, not just impressions.
The Self Financial result isn’t an outlier. It’s proof that the method works. The goal is to run the loop again, for every brand category where customers are asking questions and AI is increasingly the one answering them.
That’s the only position worth holding: being the most credible, most quotable, most trusted answer to the question your customer is already asking.
