Answer engine optimization: be the source AI cites
Answer engines don't rank pages, they pick sources. On a narrow question that source is very often a forum thread, because no brand ever wrote about it. FindMention watches 7,093 live forums and emails you when your subject comes up, so you can be the answer that gets quoted.
What is answer engine optimization?
Answer engine optimization, or AEO, is the work of becoming one of the sources an AI answer is built from. Not ranking first. Being quoted.
A search engine returns ten links and lets you choose. An answer engine writes one answer and names three to five sources. A page that is not among them gets nothing, whatever position it would have held.
So the question changes. It is no longer where you sit in a list, it is whether the model finds you when it assembles the answer.
And it finds you where your subject is already being discussed.
Where answer engines find their answers: 7,093 live forums
This is not a list. It is a registry, and it grows every day.
- Forums tracked
- over 7,000 across 14 languages
- Public profiles
- 1,890 a forum earns one at 100 discussions
- Discussions read
- over 1,572,000 and more on every pass
- Sources polled
- over 35,000 feeds, sections and pages
Nobody had catalogued most of these boards. We don't work from a list, we look for the signature of the software they run: a viewtopic.php in the address, a “powered by phpBB” in the footer. That is how a forum with twelve members ends up on the map — and a forum with twelve members is exactly where a narrow question gets a real answer.
The best covered topics
Why forums get cited on narrow questions
Ask an answer engine something broad and it quotes the same large sites as everyone else. Ask it something narrow and the list changes completely.
That second list is the one worth understanding. Answer engines do not draw from the same pool as a page of search results. They favour text where a real person works through a real case, and they reach much further down the tail than a ranking ever did.
On a narrow question the commercial web has written nothing. No manufacturer publishes about an engine shudder on a nine-year-old hybrid. The model does not have fewer sources, it has other ones, and it lands on forums.
This is structural. It will not change with the next algorithm update, because the reason is simply that nobody else wrote the answer.
Which is the good news. The places that get cited on your subject can be counted, and almost nobody is watching them.
AI visibility tools tell you the score. This tells you where to play.
A whole category of tools now checks whether ChatGPT or Perplexity mentions your brand. They are useful, and we are not one of them.
They answer with a number: you appear in eleven per cent of answers. Then comes the hard question. What do you actually do on Monday morning?
An answer is built from sources. To change the answer you have to be in a source, and on a narrow question the source is a thread where somebody asked. That is a place and a moment, not a score.
So the two go together. Measure with a visibility tool if you want the number. Come here for the threads where your question is being asked right now.
How FindMention works
- Enter a keyword
Your brand, a competitor's, a product, a question your customers ask. A word only counts when it appears whole: “gas” does not fire on “magazine”.
- Narrow it, or don't
A keyword can be limited to a country or a language, and you can rule out the words and the sites that make noise. By default we filter nothing.
- Get an email when it happens
Nothing when there is nothing, and never the same discussion twice. Every alert carries the title, the forum, the section, the keyword that fired and a direct link to the thread.
About an hour, in the median, between the post and your email. That number is the product. A thread three weeks old already has its answer, and the model has already read it. One that is an hour old is still waiting for yours.
How to get cited by ChatGPT, Perplexity and Gemini
Nobody controls what a model quotes. You control what there is to quote, and five habits separate a thread that gets used from one that does not.
- Answer the whole question
A model looks for a passage it can lift as it stands. Half an answer, or an answer that points somewhere else, gives it nothing to work with.
- Be specific and checkable
Model numbers, versions, prices, dates. A concrete detail is what makes one passage worth quoting instead of the ten vague ones around it.
- Get there early
The first useful reply in a thread is the one that gets read, quoted and built on. The fifteenth is buried, however good it is.
- Say the same thing everywhere
Models cross their sources. A claim that appears on your site, in a forum answer and in a comparison holds better than the same claim in one place.
- Say who you are
Declare yourself and answer honestly. Most forums welcome it, and a named expert is more quotable than an anonymous account, not less.
Which leaves one problem: finding the thread while it is still open. That is exactly what FindMention gives you — the timing. You show up when it counts, on the forums where people already know you.
Frequently asked questions
What is the difference between SEO and answer engine optimization?
Less than the marketing suggests, but not nothing. Both reward clear pages, real expertise and being cited elsewhere. The difference is the prize: SEO wants a rank in a list, AEO wants to be one of the three sources an answer is built from. Ten good positions are worth less than one quoted paragraph.
How do I know if ChatGPT cites my brand?
Today there is no official way. No answer engine publishes what it quoted, and none offers a report. The tools that sell this run prompts and count what comes back, which is a sample rather than a measurement. The honest method is the same one: ask the questions your customers ask, in several engines, and write down what you see.
Do forum posts really get cited by AI answers?
On broad questions, rarely. On narrow ones, constantly. Ask about a specific fault on a specific model and the sources named are threads where an owner described it. That is not a quirk of the model, it is what the model has: nobody else wrote about it.
Can you pay to be cited by an AI?
No, and that is the interesting part. There is no ad slot inside a generated answer today. What gets you in is being the source that answers the question best, which is slow — and which nobody can outbid you for.