How to Show Up in ChatGPT Answers: An AEO Playbook

Ellie at Willa, Editorial lead, Willa · Published July 15, 2026 · Updated August 25, 2026

Search is quietly moving from Google's ten blue links to a single AI-generated answer. If your brand isn't part of that answer, you're invisible to a growing share of your audience. At Willa's live session "Master AI Search: How to Show Up as ChatGPT's Answer," AI search expert Emily Richardson, who works with brands at her company Known (KNWN) after five-plus years in marketing and three years focused on AI in marketing, walked through the exact process she uses to help brands get cited in ChatGPT. This guide breaks her playbook down into steps you can run this week.

Why is showing up in ChatGPT different from ranking on Google?

The short answer, in Emily's words: "Google searches with keywords, and AI searches with math."

On Google, you type a 4 to 6 word query, Google matches keywords against an index, and you click through results and synthesize an answer yourself. AI search works differently. The average AI query is about 28 words. The engine then performs a "query fan-out," breaking your question into two to six sub-questions to get at your intent, retrieves relevant content chunks using retrieval-augmented generation (RAG), which matches vectors and embeddings rather than keywords, then filters and ranks those chunks and synthesizes one answer.

The result: AI citations don't mirror Google rankings. Emily shared an Ahrefs study of 15,000 prompts that found, on average, 81% of the web pages surfaced by large language models don't rank in Google for the same query. You can be cited by ChatGPT while ranking poorly in Google, and rank well in Google while never being cited.

The behavioral shift is real too. Emily cited figures showing that 71% of people in the US have tried AI search tools and 14% use them daily; ChatGPT alone has over 600 million monthly users. When Google shows an AI Overview, about 60% of those searches end with zero clicks to any website. And 41% of consumers say they trust AI search results more than Google, rising to roughly 68% among regular AI users.

One practical note: Emily recommends testing in ChatGPT because it's the most consumer-facing assistant, but if your customers live in Claude or Perplexity, test there. (Perplexity, she noted, aligns more closely with Google results than any other AI assistant.)

How do I pick queries I can actually win?

This was the heart of the session. Emily's biggest advice for small and mid-sized brands: be reasonable about the queries you chase. Broad queries like "best AI note-taker" are dominated by enterprise players. In her words, "you'd actually be better served picking different queries."

Start by gathering question ideas from three places: Google's "People Also Ask" boxes (questions with proven volume), asking ChatGPT or Claude what your buyer is asking (request real buyer phrasing, not keywords), and Reddit, where, as Emily put it, your buyers are complaining unfiltered.

Then understand which of her six question types you're looking at:

• Definitional ("What is an AI note-taker?"): skip these; the answer cites information, not brands.
• Procedural ("Is it legal to record job interviews?"): enterprise help centers already dominate here.
• Category ("best AI note-taker"): winnable mostly for large companies.
• Constrained category ("best AI note-taker for hiring interviews for small teams"): a top pick for smaller brands.
• Brand-anchored (questions about your brand): make sure your own site answers everything people ask about you.
• Validation ("Is [your product] worth it for a small team?"): great fodder for comparison and review pages.

For non-enterprise companies, Emily recommends focusing on the last three, and picking roughly 5 to 8 key queries across your landscape.

How do I score a query before I invest in it?

Emily scores every candidate query on four checks, each rated 0 to 2:

1. Does the answer name anybody? Run the query in a temporary chat. If the response is a generic definition with no brands cited, walk away. "If the answer does not name anybody, nobody's gonna win, including you."
2. Foothold: Is any company your size already in the answer? If only nine-figure-revenue brands appear, that's a bad sign; if a peer got in, that's a great one.
3. Specificity: Does narrowing the question change who shows up? Her rule: "If narrowing the question gives you a different answer, you're looking at the right question." In her live demo, "best AI note-taker" surfaced Granola, Fathom, Otter, and Fireflies. Adding "for hiring interviews" surfaced a different set of niche tools, which signals an open lane.
4. Sources: Who gets cited, and can you get in? If citations are Wikipedia, Gartner, major newspapers, or .gov sites, score it low. If they're G2, Reddit, Substack, listicles, and normal company websites, that's reachable.

Total the score: 6 to 8 means pursue the query now; 4 to 5 means only if it's core to what you sell; 0 to 3 means walk away.

Once you've picked a query, ask ChatGPT for five other ways someone might phrase the same question. That's your "query cluster": the set of related phrasings you'll create content around and track together, which matters because AI search fans one question out into many.

How do I track whether ChatGPT mentions my brand?

You don't need paid tooling to start. Emily's manual process:

1. On a set cadence (weekly, biweekly, or monthly), open a temporary chat, ideally logged out or in a separate browser, so ChatGPT's memory of you doesn't bias the results.
2. Run each query in your cluster.
3. Copy every source cited into a spreadsheet, and copy the answer text itself.
4. Log: Were you named? Who else was? Did the phrasing of the answer change?

Run important queries a few times, since results vary, and treat this as directional, not exact science. As you publish and update content, this log shows whether it's moving the needle. Two extra tips from the Q&A: add "AI search" as an option on your "How did you find us?" form, and use bot-tracking (Google Analytics now tracks AI bots) to see which pages AI crawlers visit. Rising bot traffic on a topic is a signal to double down.

If you'd rather automate, Emily mentioned analytics platforms including her own company, Known, and Peec AI, which she noted is tailored to smaller brands.

How do I optimize my content to get cited?

Emily's core principle: "You are not going to get cited in AI search if it doesn't actually have to search." Generic content answering generic questions gets answered from the model's memory. Content that combines niche topics with unique expertise or a distinct point of view forces the engine to go looking. That's when you can be the citation.

Her most concrete, do-this-now tactics from the session:

• Put the question in your header. A question-phrased heading makes you about 30% more likely to show up as a citation.
• Front-load the answer. Put a summary directly under the header. About 70% of the time, the AI reads only the top third of your page.
• Publish listicles and honest comparison pages. Roughly 50% of B2B citations are listicles; AI would rather cite one page that already did the comparison than assemble it from many pages. Don't just rank yourself first. Genuinely credit competitors' strengths.
• Append FAQs. Run your existing content through AI to generate FAQs and add them to the page.
• Avoid JavaScript-only content. If your content isn't server-side rendered, AI crawlers largely can't read it, including FAQ dropdowns built in JavaScript.
• Make your help center public. Emily called moving documentation out from behind logins one of the biggest wins for lower-funnel visibility, along with dedicated integration pages.
• Keep your brand description identical everywhere. Because AI matches vectors, using the same name and tagline across your site, LinkedIn, G2, and other profiles builds the engine's confidence in what you are.
• Publish opinions. "The AI can't have experiences, and it does not form very strong opinions. It only gathers opinions," Emily explained. A strong point of view on Substack, Medium, or LinkedIn is exactly the kind of thing AI pulls in and attributes. And if you want to show up in Gemini, make YouTube content: YouTube is Google.

Finally, reverse-engineer the winners. Paste the sources ChatGPT cited for your target query into a new chat and ask it to identify the listicles you could pitch your way into, the formats you're missing (comparison pages, pricing breakdowns), and why each winning page won. Then build your versions.

Key takeaways

• AI search works mechanically differently from Google: 81% of LLM-cited pages don't rank in Google for the same query, so you need a distinct strategy.
• Small brands win on specific, constrained queries, not broad category terms. Pick 5 to 8 and score each on naming, foothold, specificity, and sources before investing.
• Track a cluster of 5 to 6 phrasings per query weekly in a temporary, logged-out chat, and log every citation in a spreadsheet.
• Question-phrased headers, front-loaded answers, honest comparison pages, public help centers, FAQs, and server-rendered content are today's highest-impact optimizations.
• Opinionated, niche content gets cited; generic content gets answered from memory. Publish your point of view.

This session built on skills many Willa members are already developing. If you're earlier in your AI journey, start with AI Skills for Non-Technical Professionals: Where to Start, and when your AI search wins start compounding, add them to your resume.

Join us for the next one

This walkthrough came from a live Willa event, where members watched Emily run every step on screen and asked their own questions, and she's hinted at a part three. Willa hosts expert-led sessions like this almost every week for women building careers with AI. See what's coming up on our events page, or become a member and join the community that's learning this stuff together, in real time.

• AI search
• AEO
• ChatGPT
• SEO
• marketing

Join Willa, the community where women learn AI together.

All posts