What is Answer Engine Optimization (AEO)?
Answer Engine Optimization is the practice of structuring your site and your wider web presence so that AI assistants — ChatGPT, Perplexity, Google's AI Overviews, Copilot — cite your business when someone asks a question you can answer. It matters because ChatGPT alone drives roughly 87% of AI referral traffic, and that traffic converts several times better than ordinary search. It is worth paying for only if the agency can actually measure whether you are being cited, which most cannot.
Why this appeared so suddenly
Search behaviour changed faster than the agency industry did. People increasingly ask an assistant for a recommendation instead of scrolling a results page, and the assistant answers with a handful of named businesses. If you are not one of them, the visit never happens — there is no page two to be on.
| Signal | Figure |
|---|---|
| ChatGPT's share of all AI referral traffic | ~87% |
| Growth in AI referral traffic, 2024 → 2025 | +527% |
| AI traffic vs organic conversion, median across 78 sites | 1.26× |
| AI traffic as a share of all sessions | 0.2% – 3% |
| Pages with schema markup — likelihood of being cited | 78% higher |
| Citations coming from third-party sources vs your own site | 6.5× more likely |
| Marketers calling AI optimization a core 2026 strategy | 43% |
| Marketers actually measuring AI citations | 14% |
Two honest caveats, because you will see wilder numbers quoted. The often-repeated claim that AI traffic converts 4× to 23× better than organic comes from single-company funnels; the only multi-site study that screened its sample and controlled for outliers found a median of 1.26×. And AI is still only 0.2–3% of sessions. It is growing fast and it converts somewhat better — it is not yet a flood.
The last two rows of the table are the real commercial story. Three times as many people say this matters as have any way of knowing whether it is working for them.
What actually drives a citation
AEO overlaps with SEO but is not the same job. Ranking is about being the best result for a query; being cited is about being the most quotable source for an answer. In practice that means:
- Schema markup on every page — the single clearest technical signal, and pages that carry it are 78% more likely to be cited
- Question-shaped headings with the answer in the first two or three sentences, because that passage is what gets extracted
- Concrete numbers, tables and dates rather than adjectives — assistants quote specifics and skip marketing language
- Third-party mentions, which drive citations 6.5× more than your own site does; listicles, directories and LinkedIn matter more than another blog post
- An llms.txt file pointing assistants at your most useful pages
- Crawler access actually allowed — a surprising number of sites block GPTBot by default and never notice
That third-party figure is the one most agencies get wrong. If someone proposes an AEO programme consisting entirely of on-site work, they have misread where citations come from.
What it costs
| Tier | Monthly |
|---|---|
| Small-business AEO, visible market floor | ~$797 |
| Typical B2B engagement | $2,000 – $10,000 |
| Enterprise programmes | $25,000 – $50,000+ |
| One-off AEO audit | $1,000 – $4,000 |
| Our Visibility plan | $600 |
| Our one-off audit | $1,000 |
There is no established price anchor at the small-business end yet, which is exactly why quoted prices vary by an order of magnitude for work described in identical language.
How to tell whether an agency can really do this
Ask one question: "Can you show me, today, whether ChatGPT currently mentions my business — and can you show me that again next month?" It is a hard question to bluff. Measuring it means repeatedly querying the assistants, recording what came back, and tracking the change over time.
- If the answer is a slide about "AI-first content strategy", they are selling SEO with a new label
- If the answer is a screenshot of one lucky query, they have no measurement, just an anecdote
- If they can hand you the tracked queries and the history, they have built something
Is it worth it?
Honestly: for some businesses, not yet. If almost nobody asks an assistant for what you sell, spend the money elsewhere and revisit in a year. AEO earns its place when your buyers research before choosing — professional services, considered purchases, anything where someone asks "who's good at X near me" before picking up the phone.
What is not defensible in 2026 is paying for visibility work that nobody measures. That was tolerable when the only channel was Google and rankings were at least checkable. It is not tolerable now that a growing share of demand is decided inside a conversation you cannot see.