Answer engine optimization is the practice of writing content so that AI tools like ChatGPT, Perplexity and Google AI Overviews can lift a finished answer out of it and say where they got it. The basics are not that different from good SEO, but the prize has changed: you are no longer competing for a click, you are competing for one sentence inside somebody else's answer. Here is what AI search actually cites, how a small company site can prepare for it, and a twenty minute test that tells you whether AI knows you exist.
Key takeaways
- Answer engine optimization means structuring pages so a language model can extract a complete, attributable answer without guessing.
- Ahrefs compared 300,000 keywords between March 2024 and March 2025 and found that the click through rate of the top organic result dropped by 34.5 percent when an AI answer appeared above it.
- AI answers are no longer a US only story. They rolled out across Europe, including small markets like Slovakia, during 2025.
- Models cite pages with a direct answer in the first sentence, clean structure and verifiable numbers. Brochure language gets ignored.
- Ranking well still matters. AEO sits on top of SEO, it does not replace it.
- The twenty minute test below uses five prompts and tells you where you stand today.
What is answer engine optimization and how is it different from SEO?
Answer engine optimization is preparing your content to be used as a source inside a generated answer. SEO fights for a position in a list, AEO fights for a mention in the paragraph people read instead of that list. The goal differs, the craft overlaps: without solid content and a technically healthy site, AEO has nothing to work with.
You can see the difference in how buyers behave. Someone asking what a corporate video costs used to open three agency sites and compare. Now the assistant gives them a range, sometimes with sources. If you are in that answer, you earned trust before the first click. If you are not, that person never learns you exist, even if you sit second in Google.
So stop thinking in keywords and start thinking in questions. A question is the unit an AI works with.
How much traffic do AI answers actually take away?
Enough to notice. In their study of answer engine optimization basics and click behaviour, Ahrefs compared 300,000 keywords between March 2024 and March 2025 and measured a 34.5 percent drop in click through rate for the number one organic result once an AI Overview appeared above it. Search is not dead, but the same position brings fewer people than it did two years ago.
We see the same pattern on our own site. On pages about production pricing, fewer visitors arrive by scanning a results list and more arrive already holding a summary an assistant gave them. The first contact with the brand moved into an answer we do not control.
That is the honest reason to care. Not because AEO is fashionable, but because the free traffic you built over years now passes through a filter that quotes only some of it.
How do you write a page an AI will pick?
The answer has to be written out, not implied. A model will not assemble your conclusion from three paragraphs of build up. When we write for clients, we hold to four rules:
- Question as the heading, answer in the first sentence below it. Not "Our approach to pricing" but "How much does a corporate video cost?" followed immediately by a range in euros.
- Numbers with the source inside the sentence. "According to a 2025 Ahrefs study" reads very differently to a model than "studies show".
- Short, self contained blocks. Bullet lists, comparison tables and an FAQ section. Those are the parts that get quoted most often.
- Specifics nobody else can publish. Your real prices, your real process, your real mistakes. Generic advice exists in a thousand places, so there is no reason to cite yours.
Add the technical baseline that teams keep forgetting: pages that load fast and can be read without javascript, meaningful headings, and schema markup for questions. If you are commissioning a new site or a content strategy, raise this at the start. Retrofitting it is expensive.
What does not work in AEO?
A fair amount of what gets sold as AEO today is a rebranded content package. Three things we do not believe in.
Publishing more. Fifty average articles do not make a model quote you more often. In July 2026 we cut our own blog from twelve pieces a week to seven for exactly that reason, and put the time into depth instead.
Bolting an FAQ block onto every page. If the answers say nothing, it is filler with a schema wrapper.
Precise measurement. No tool reliably tells you how much revenue came from an AI answer, so most AEO dashboards are estimates dressed up as data. We keep a plain list of prompts and check them by hand once a month. Primitive, but honest.
How do you test whether AI knows your company?
Do this yourself, no tool required. Open ChatGPT or Perplexity and run five prompts:
- "Who are the best companies for [your service] in [your country]?"
- "How much does [your service] cost in [your country]?"
- "How do I choose a supplier for [your service]?"
- "What does [your company name] do?"
- "What are the alternatives to [your biggest competitor]?"
Write down whether you appear, who appears instead of you, and which pages the model pulled from. If you are missing even from the prompt with your own name, there is not enough clear public content about what you do. If the description is wrong, your site contradicts itself somewhere. If you are absent from the pricing prompt, you have published nothing about pricing, which is the most common case by far.
Put it in a spreadsheet and repeat it in a month. You will see movement long before any report tells you about it.
Is this worth it for a small business?
If people research your product before buying, yes. The more expensive and unusual the service, the more questions a buyer asks, and the higher the chance an assistant answers first. For impulse purchases or purely local services it is still marginal, and skipping it is a reasonable call.
The good news is that AEO does not need a new budget. It needs the content you already produce, written differently. If you want a second pair of eyes, get in touch and we will go through your prompts and what AI currently says about you. If you are only starting with content, begin with a marketing strategy instead, because optimising an empty site optimises nothing.
Frequently asked questions
Will AEO replace SEO?
No. Both Google AI Overviews and ChatGPT lean on content that already performs somewhere, so a technically healthy site and decent rankings remain the foundation. AEO is an extra layer, not a swap.
How long does it take to show up in AI answers?
From our own projects and client sites, expect weeks to months after a page with a direct answer goes live. Narrow topics move faster, simply because the model has less to choose from.
Does a mention help if nobody clicks it?
Yes, but it shows up differently in your numbers. A mention works like a recommendation and people look you up by name later, so track direct visits and branded searches, not only organic clicks.
Do we have to rewrite the whole website?
No. Start with the five pages that matter most, usually pricing, your main service page and two or three articles. Turn the headings into questions and the opening lines into direct answers. The rest can wait.
