There’s a growing wave of bad advice telling founders to optimize for ChatGPT.
The pitch they’ve heard:
There’s a new SEO in town—Answer Engine Optimization (AEO)—and you’d better pour budget into it to be featured in LLM answers.
I’m writing this article to give a sober assessment of AEO for early-stage founders and marketing teams.
What’s real, what’s noise, and where the ROI actually lives?
The curiosity is valid. The return usually isn’t.
For most companies, AEO won’t drive meaningful ROI. Not because the concept is flawed, but because the opportunity is too narrow to move the needle.
It’s easy to assume that as AI-generated answers become more common, being “featured” in them must be valuable.
But the numbers tell a different story.
ChatGPT search feels everywhere and it is growing fast.
Does anyone still you Google for search in 2025?
Short answer is this:
While ChatGPT's usage grows, its scale remains modest compared to Google.
Also, the perception that Google is losing search traffic is not supported by data. In fact, Google's search volume experienced a 21.6% increase in 2024, and its market share remains above 90%.
On Google, nearly every visit is an explicit hunt for information, navigation, or a purchase.
On platforms like Google, the distribution of queries has historically followed a predictable pattern: roughly 50% of searches are informational (“what is”, “how to”), about 30% are navigational (searches for a specific brand or URL), and the remaining 20% are transactional or commercial (“buy”, “best X”, “pricing”).
But when we look at ChatGPT, the pattern shifts dramatically.
| Query Type | Google Search (~) | ChatGPT (~) | Example |
|---|---|---|---|
| Informational Fact-based queries, how-tos, general knowledge |
50%+ | 15% | “what is AEO?”, “how to speed up CI/CD” |
| Navigational Looking for a specific site, product, or brand |
~30% | 10% | “hubspot login”, “open stripe dashboard” |
| Transactional / Commercial Buying, comparing, or pricing products/services |
~20% | Minimal | “best project management tools”, “buy laptop” |
| Creative / Generative / Assistive Writing help, code generation, brainstorming, analysis |
Rare | 75% | “write a cold email”, “explain TCP/IP to a 12-year-old” |
Only about 15% of its prompts are actual classic informational search.
The vast majority of usage, nearly 75% consists of tasks that traditional search engines were never designed to handle. These include writing assistance, ideation, coding, summarization, and creative synthesis.
Implication: the addressable “search‑like” tasks inside LLMs is small. Unless your business lives in those informational questions, an AEO mention is unlikely to deliver meaningful traffic or leads.
AI-generated answers often satisfy the user immediately on the spot. Once the question is solved, most people stop there.
Implication: (AEO) impact is often limited even when your content is referenced.
Yes, a mention in AI-generated answers can boost credibility, but it rarely drives measurable traffic or sales. Most companies should treat AEO as a lightweight brand play, not a growth channel.
Chasing LLM citations is almost never a positive-ROI move—but search behaviour is clearly evolving.
Queries are getting longer and more conversational, link-hopping sessions are shrinking, and engines now reward tight, self-contained explanations instead of long copy.
| Traditional Search Model | Now: Answer Engine Model | |
|---|---|---|
| Query Format | Short, keyword-based phrases | Full, natural language questions |
| Example | best productivity tools 2025 |
“What are the best productivity tools for remote teams in 2025?” |
| User Behavior | Try multiple keyword combinations and click through links | Ask one clear question and expect a complete answer |
| Click Intent | Click required to access and evaluate content | Often no click — answer is displayed directly |
| Interface Expectation | List of ranked links (SERP) | AI-generated summaries, direct answer box |
| Role of Search Engine | Navigation tool to find information | Answer agent that synthesizes and delivers it |
| Content Relevance | Driven by ranking, backlinks, authority | Driven by clarity, structure, semantic relevance, trust |
That shift does have strategic implications for founders and marketing teams. Instead of pouring money into speculative “LLM PR,” apply a pragmatic AEO layer to the content you’re already producing.
Here´s how.
Think of AEO like writing unit tests for your content.
A good test case has a clear input and an expected output. So does answer-optimized content.
Q: What is Answer Engine Optimization?
A: Answer Engine Optimization (AEO) is the practice of formatting content so that AI search engines and chatbots can directly extract and display it as a concise, trustworthy answer to a user’s question.
Use this pattern across your site — in FAQs, blog posts, documentation.
Use tools like Google’s "People Also Ask," and Reddit to surface how people actually ask.
AEO strategy starts with question-first thinking.
When someone asks, "What is AEO?" they’re not looking for an anecdote. They’re looking for a 1–2 sentence definition.
Give that answer first. Then elaborate. Use the inverted pyramid—lead with value, add nuance later.
The inverted pyramid is a metaphor used by journalists and other writers to illustrate how information should be prioritised and structured in prose (e.g., a news report). It is a common method for writing news stories and has wide adaptability to other kinds of texts, such as blogs, editorial columns and marketing factsheets.
It’s worked for people since 1845—and it works for LLMs in 2025.
Your content should be:
This isn’t about tricking models.
It’s about making your content easier to parse—for readers and for systems.
The easier you are to extract, the more likely you’ll be cited.
LLMs weigh recency and trust. So do users.
Fresh, specific, human-readable content gets cited more by people and machines alike.
Help LLMs read your content with llm.txt
Think of llms.txt as robots.txt for LLMs. It is like a tiny “welcome note” you leave in your website’s front window—written for AI assistants instead of human visitors. It points large-language-model crawlers to the pages you’re proud of and away from the bits you’d rather keep private.
Like adding alt-text to images: tiny effort, possible upside, zero downside.
Search behavior has changed — fast. But the future of content isn’t about gaming your way into LLM answer boxes.
Answer Engine Optimization (AEO) is a side effect—not the goal—of great content.
Don’t optimize for the model.
Optimize for the person with the question.
The best way to get surfaced by a machine is to write like a human.
Rafael Pinheiro
Co-founder at Ottic