How to Build Your Brand's AEO Strategy
How LLMs are changing the way we brand-build online, 3 companies doing it well, 5 things to do now, and an expert Q&A
I recently spent an afternoon with 150 other marketers at AEO Conf, a one-day conference in San Francisco focused on “answer engine optimization” (with 1,300 people on the waitlist—clearly I’m not the only one eager to figure this out). The event was hosted by Graphite, AirOps, and Webflow at The Conservatory, and if you’re not already tracking what’s happening with AEO, its time to start. LLMs are now telling your brand story whether you like it or not—AEO is how you control the narrative.
P.S. Had a blast meeting some On Brand readers IRL at Tallboy the other night! Nothing beats face to face—it’s the perfect antidote for when you’ve been thinking about AI a little too much (yes, it’s fun and exciting when you make a breakthrough with Claude Code, but too much AI news makes you want to go full luddite…it’s a daily struggle).
If someone sent you this post and you're not subscribed, join those people learning how to tactically advocate for brand at your company. 📬
This article is a deep dive, so let me give you a preview of what’s ahead:
How AI is now part of how your brand story gets told
How people’s behavior is changing around search and chat
5 things you can do right now + 3 companies doing it well
My follow-up conversation with Ethan
What about all the slop?
AI is building your brand story from consensus
Your brand has always existed in places you don’t fully control—in customer service calls, in Reddit threads, in offhand comments on Instagram. In the “old world,” those mentions were fragmented. Someone complaining about your product on Reddit would be seen by a few hundred people, maybe a few thousand if it went viral.
In the AI world, that Reddit comment gets synthesized alongside every mention of your product. Forever. Or at least until people post something different enough times that the “consensus” shifts.
As Ethan Smith, CEO of Graphite, put it in a conversation we had after the conference:
“AI is now telling your brand story. You need to be part of that conversation so you don’t lose control of your brand narrative.”
AI is forming audience perception by synthesizing everything it can find about your brand—on your website site, yes, but also across social threads, YouTube videos, review sites, LinkedIn posts, anything public. Every mention becomes part of your brand story, whether you participated in creating it or not.
A slide from the opening session:
This is the ultimate extension of what I’ve been saying forever: you don’t control your brand. (You control your branding and your brand marketing. In one of my first posts, I broke down the difference between brand, branding, and brand marketing, all of which get conflated too often.)
Your brand exists in every customer service interaction, every error message, every Glassdoor review—all of those are forming your brand whether you’re actively shaping it or not.
But now, the stakes are higher because AI is actively building a consensus view of your brand from every mention it can find online.
The numbers are bigger than you think
Let’s start by talking about how user behavior and expectations are changing. Search-related use of AI accounts for almost a third of all search traffic, with 75% happening on mobile. AI (LLMs) receive 5.4 billion monthly sessions in the US and 45 billion monthly sessions worldwide, and December 2025 saw a 300% year-over-year (YoY) increase in AI usage in the US (source).
While traditional search has been dominated by short, “head” keywords that drive most traffic, AI search is almost entirely long-tail. In traditional Google search, 1-3 word queries dominate—26% are 2-word queries alone. But in AI? 60% of prompts are 10+ words, while nearly 30% of all prompts are 21+ words long.
Google itself is obviously evolving to be AI-first—1.5 billion people are now seeing Google’s generative results, and the prevalence of AI Overviews in search results is surging while Featured Snippets are declining.
We know this, too, is changing behavior; for example, it affects click-through—Google users who encounter an AI summary are less likely to click on links to other websites than users who do not see one, and seeing an AI summary is more likely to end their search.
There’s no doubt more deeply embedded AI experiences are around the corner. Your growth strategy has to work across both “old search” and “AI search.”
I’ve never considered myself an SEO nerd, but I’m paying attention because this isn’t a change in channel strategy. It’s a change in how people ask questions and find answers—way bigger deal for how people engage with the internet.
More “asking” is happening now
One of the most compelling insights from the conference was that the total market for “asking” is expanding. Rather than replacing traditional search, AI is creating new behavior.
People are asking questions they never would have typed into Google. They’re having conversations with AI (long ones, at that) because it feels more like consulting a colleague than querying a database. And unlike previous technology shifts that took years to reach critical mass, buyers are moving decision making to this AI channel faster than we’ve ever seen with any new tech.
At least 800 million people now are using ChatGPT. (As of the end of last year, ChatGPT owned 86% of the US AI market share.) Since it launched in November 2022, total usage of search (combining traditional search engines and AI “asking” prompts) has increased by 26% worldwide and 16% in the US.
Search traffic hasn’t declined, it’s grown.
But, for the first time in over a decade, Google’s market share has declined. Worldwide, Google’s share decreased from 89% in 2023 to 71% in Q4 2025. ChatGPT now commands 20% of search traffic worldwide (12% in the US). But Google’s absolute traffic hasn’t dropped; again, the pie just got bigger.
Where AI is looking (and what it’s finding)
Later in the day, Rob Gaige, Reddit’s Global Head of Insights, and Mayank Yadav, Director of Product at Reddit, broke down how this consensus-building plays out—how what you see in those AI summaries or in LLM responses ends up there. Reddit now has 450 million weekly active users globally (up 24% YoY), with massive reach in key markets. And posts with commercial intent on Reddit are up 20% YoY.
Unfiltered Reddit conversations—the ones where someone asks “is X actually worth it?” or “what’s the best alternative to Y?”—are revealing real customer intent, pain points, and buying signals. And AI is scraping all of it to form a narrative about your brand.
Citations are super fragmented across sources. Even the top-cited domain (reddit.com) only accounts for 2.36% of all citations. The distribution falls off quickly from there—youtube.com at 1.89%, wikipedia.org at 0.88%, linkedin.com at 0.85%.
Start by searching for your brand, your product, your category, and your competitors across these sites. Not to see if you can game the system—you can’t, and I don’t recommend you try. But to understand what narrative is being formed about you, and where the gaps are that you can fill with authoritative, helpful content.
Apt moment for another reminder that branding is virtually free to invest in. Your homepage, pricing page, customer support scripts, error messages, careers page, social media presence, candidate follow-up experience—all of these are opportunities to shape perception with intention. The only difference now is that AI is watching all of it, forming a “consensus,” a brand narrative, from every signal you put out there…and the ones everyone else does, too.
Inconsistencies across your collateral weaken those trust signals. Tech changes, story stays the same—if you aren’t ahead of crafting your narrative, others will do it for you. And you might not like where it ends up.
The long tail is where you win
Most marketing stacks were designed for a world where you could control the message through owned channels and paid placement. Traditional search taught us to prioritize everything around high-volume keywords. The best SEO teams got really good at ranking for those money terms that drove most of the traffic. But in AI search, those head terms matter less. The long tail is everything.
Ethan put it perfectly:
“Prompt data is the new keyword data.”
Where do you find these long-tail queries? Not in your paid search dashboard. You find them in:
Sales call transcripts
Customer support tickets
Community forums and Reddit threads
User interviews
Help desk conversations
These are the questions people are already asking somewhere—just not necessarily in Google. They’re asking them on calls. They’re messaging them to peers. And when you can answer a 20+ word prompt that nobody else has addressed, you win the citation.
The mechanics are similar to traditional SEO, but the opportunity is bigger. Much bigger. Because while only 5% of your landing pages probably drive 86% of your traditional search traffic, AI can surface that “wasted” content if it answers specific, detailed questions that matter to your audience.
Not all AI usage is search, of course. I’ll go on a quick tangent here to include that about 52% of AI usage is “Asking” prompts that compete with traditional search. The other 48% is split between “Doing” prompts (35%)—like “rewrite this email to make it more formal”—and “Expressing” prompts (14%)—statements that aren’t requests for information or tasks (“I’m worried about this situation”).
So when comparing AI to traditional search, we should really only count those “Asking” prompts. Which means search-related AI usage is currently 17% the size of traditional search in the US (28% worldwide). Still massive, still growing at 300% year-over-year, but a more accurate comparison.
5 things to do right now
To start on how brand leaders should think about this change and opportunity, I had another chat with Ethan after the event:
KK: What’s the first place marketers should look to understand what story AI is telling about them?
ES: I think you can just ask AI, “Tell me what you know about [my brand]?” Then ask other questions like “Help me decide between [my brand] and [my competitor].” “What are the most common complaints about [my brand]?”
KK: Most brand teams are used to controlling narrative through owned channels—our website, our ads, our content. But you showed that citations are fragmented across dozens of domains. How should teams think about resource allocation when they can’t be everywhere?
ES: It’s impossible for someone to directly optimize and build programs for hundreds of thousands of sites. Instead, I would focus on offsite owned channels (my profile on other sites like YouTube and X), content marketing and social media (e.g. viral content that your brand creates that is shared everywhere, e.g. https://www.axios.com/2025/10/14/ai-generated-writing-humans ), PR, and affiliates. I think PR in particular is going to have a “PR + AI Renaissance” because PR is the best way to get hundreds of mentions at scale.
KK: As someone leading PR right now, I’d have to agree! Okay, the data shows that 60% of AI prompts are 10+ words, and people are asking questions they never would have typed into Google. This is a big shift in search, but it also feels like a massive shift just in how people expect to interface with brands. If you can imagine this far ahead… five years from now, what does “brand building” look like in a world where brand discovery happens through AI conversation rather than search results or social feeds?
ES: I actually do not think that brand building will happen mostly in AI, and not in search and social media. I think all three will continue to be the places that your brand story is told. You will tell your brand story in three places 1. Owned (my site and products) 2. Earned (social media) and 3. AI (ChatGPT). [Author’s note: Collab framework to come??]
KK: Building on that—I think we’d agree that in-person, offline, physical touchpoints (events, experiential, OOH, etc.) will continue to be major differentiators. Take AEO Conf! That stuff will never not matter, unless we truly enter Black Mirror and all move our brains to the cloud or whatever. 🤷♀️
When Ethan walked through immediate action items, I was practically transcribing his talk word-for-word (you’re welcome)—here’s the playbook for 2026:
Pick your top prompts. Start with your most valuable paid search keywords and your top sales questions. Ask Claude, Perplxity, or ChatGPT to find threads about your product and your competitors—what are people actually asking? Count the comments and engagements. Now you have a prioritized list of questions to answer.
Track your top-cited domains. The domains that matter to you will vary by category. YouTube might be crucial for some businesses, irrelevant for others. Look at what’s actually getting cited in AI responses about your space. Pretend you’re a prospect yourself and see what answers LLMs give you!
Build on-site content. Your website is the foundation. Help centers, tutorials, academies, integration pages, comparison content—answer those long-tail questions with depth and specificity. If you can answer a question nobody else has, you can win.
Create off-site owned content. Make a YouTube channel. Answer the questions people have about what your product does and what problems it solves. Then distribute that same content across LinkedIn, Instagram, TikTok. AI summarizes for consensus, so the more places you show up saying the same thing, the better.
Accept what you can’t control (but pay attention to it). Off-site earned content—Reddit comments, reviews, random mentions—is nearly impossible to manage directly. Don’t waste energy trying to manipulate these channels or delete negative comments. But do pay attention to what’s being said. Those Reddit conversations reveal real customer intent and pain points. Use them to inform your content strategy.
The most useful part of any conference is seeing what’s actually working. A couple examples stood out:
Klaviyo created a comprehensive integration guide that answers every question about connecting to Shopify, from basic setup to advanced customizations. Instead of one page covering everything, they built separate pages for each specific question. They’re now cited in 80% of AI responses when someone asks about email marketing for Shopify.
Zola rebuilt their help center around actual customer questions pulled from support tickets and Reddit threads. Each article targets one specific long-tail query. They went from almost no AI citations six months ago to being mentioned in 40% of responses about wedding registries.
Assembly AI published detailed comparison content for every major speech-to-text API, including their own. They didn’t hide/gloss over competitor advantages, instead they were honest about trade-offs. AI started citing them as the definitive comparison source, which drove more qualified leads than any of their traditional SEO content.
The pattern, for your content & social strategy: They stopped trying to rank for broad terms and started answering specific questions better than anyone else.
What about all the slop?
One question that came up multiple times: How do you maintain quality when you’re creating more content across more channels?
Quantity is not the answer to a great AEO strategy. The answer is in how you design a system for it. If you start with a clear and aligned context layer, if you have the right checkpoints for human review, if you’re grounding your content in actual customer conversations and technical accuracy, these will all help you avoid the generic AI-generated dreck that’s flooding the internet.
That part hasn’t changed. You still need to execute with taste. Yes, “taste” has become this weird overused buzzword (since I first wrote about it in 2024) but I’m still arguing that taste is a learnable skill that in the context of work heavily translates to empathy for your user.
Practicing taste means being deliberate about when to let automation run and when a human needs to weigh in. And that’s key to building flywheels where customer conversations inform content strategy, which improves visibility, which brings in more customers whose questions feed back into the system.
Building a resilient team
Silicon Valley loves to sell bold visions of transformation and knowing what’s happening at the cutting edge. But there was consensus (pun unintended) in that room that you can’t control how everything unfolds. New models can completely shift AI search behavior overnight. Sources and domains explode in importance and then fade. The landscape is volatile, to say the least.
What you can control is building a resilient team, building a brand-led culture. A team that learns fast. A team that treats this like a system-building challenge, rather than a tactics problem.
The marketers who can figure out how to build these systems—how to gather intelligence, centralize context, design effective operating models, and execute with quality at scale—are the ones who’ll be running brand and growth in a few years.
The rest will still be Googling “how to do SEO.” 🍞
Thanks for reading.
If you’re thinking about AEO strategy at your company, I’d love to hear what's working (or what you’re struggling with)—or check out my friends at Graphite or AirOps. Hit reply or drop a comment!
And if this was helpful, consider:
replying and saying hi!
connecting with me (Kira Klaas 👩🏼💻) on LinkedIn
taking the 2-minute reader survey ‼️
sending to a friend 💌 or coworker 💬
Thank you. This is very helpful!
Very helpful. I'll definitely be coming back to this article more than once.