Being Visible in AI Search: An AIO / GEO Guide
Search is no longer just ten blue links. A model answers the user's question and points to a few sources. The whole game is being one of those sources.
The goal of classic SEO was ranking. The new goal is being quoted. Users now ask ChatGPT, Perplexity or Google's AI Overviews box, and a few sources appear under the answer. This piece covers the concrete steps to being one of them.
What are AIO and GEO?
AIO (AI Optimization) or GEO (Generative Engine Optimization) is the work of making your content readable, understandable and quotable by AI-powered answer engines. It does not replace classic SEO; it sits on top of it.
In SEO the goal was to be clicked. In AIO the goal is for your name to appear inside the answer, even if nobody clicks.
1. Technical ground: the model has to be able to read you
- Server-side rendering. The crawlers of answer engines mostly do not run JavaScript. The content has to be visible in the first HTML.
- Allow AI crawlers. GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot and Google-Extended should not be blocked in your robots.txt.
- llms.txt. A plain-text file summarising what the site is and what its important pages say. It helps models grasp the site quickly.
- Clean page structure. One H1, a sensible H2/H3 hierarchy, short paragraphs.
- Speed. Slow pages eat both the user and the crawl budget.
2. Structured data: a sentence the machine can read
schema.org markup added with JSON-LD stops the page's subject being a guess. The ones that pay off most:
| Schema | Where | What it provides |
|---|---|---|
| Organization | Whole site | Brand entity, contact, social accounts |
| BreadcrumbList | Sub-pages | Understanding of the site hierarchy |
| FAQPage | FAQ sections | Direct quoting of question-answer pairs |
| BlogPosting | Articles | Author, date, topic and freshness signal |
| Service / Offer | Service pages | A clear definition of what you sell |
3. Writing quotable content
Answer engines do not quote long, winding text; they take clear sentences that stand on their own. The writing style that works in practice:
- Make the question a heading and give the answer in the first sentence. Do not write a warm-up paragraph.
- Give numbers and ranges. Not “fast”, but “typically 8–14 weeks”.
- Close definitions in a single sentence. “RAG is the architecture that has the model find the relevant sections of your own data before producing an answer.”
- Put a date on it. Content with a publication and update date stands out on questions that need freshness.
- Use lists and tables. Content with visible structure is easier to parse.
4. Entity signals
Models recognise brands in context. The same name, the same description and the same contact details should be consistent on your site, on your social profiles, in directories and wherever you are mentioned. A brand name written differently in different places cannot be merged into a single entity.
5. Measurement: how do you know?
- Server logs. Track visits from GPTBot, ClaudeBot and PerplexityBot; that is the most direct signal.
- Referral traffic. In your analytics, look at sessions coming from sources like chatgpt.com and perplexity.ai.
- Manual tests. Ask your target questions to different answer engines at regular intervals and record who gets quoted.
- Brand searches. Appearing in AI answers often comes back as a rise in brand searches rather than a direct click.
The difference between classic SEO and AIO
| Topic | Classic SEO | AIO / GEO |
|---|---|---|
| Goal | A top position in the ranking | Being cited as a source in the answer |
| Measure of success | Clicks, position | Being quoted, brand recall |
| Content format | Long, comprehensive page | Clear sections that stand on their own |
| Keywords | Term-focused | Question and intent-focused |
| Technical priority | Crawlability, speed | Content in the first HTML, bot permissions, structured data |
| Freshness | Useful | Critical — dated content stands out |
The two lists overlap heavily, which is good news. Almost everything done for AIO also works in classic search. The reverse is not always true: a page written for keyword density with filler paragraphs can hold on in a search engine, but an answer engine will not quote from it.
Update rhythm: publish and forget does not work
Answer engines use freshness as a strong signal, especially on questions like “how much in 2026” or “which one is better right now”. A practical arrangement:
- Review articles containing numbers twice a year. Prices, timelines and version numbers are the first things to go stale.
- Write the update date visibly and update the `dateModified` field in your structured data as well.
- Change what has genuinely changed. Stamping a new date without touching the content loses trust over time.
- Growing an existing article usually pays better than writing a new one. Opening a second page on the same topic means competing with yourself.
Checklist
- Server-rendered content, visible in the first HTML
- Explicit permission for AI crawlers in robots.txt, plus a sitemap declaration
- An /llms.txt file and an up-to-date XML sitemap
- A self-referencing canonical on every page
- Organization, BreadcrumbList, FAQPage and BlogPosting schemas
- One H1 per page and a consistent heading hierarchy
- Sections with question headings that answer in the first sentence
- Visible publication and update dates
- A consistent brand name, description and contact details
- An RSS feed and a content update rhythm
Every item above is applied on this site; you can look at the /llms.txt file to see how it looks. If you would like the same setup built for your own site, write to us here.
Frequently asked questions
What is AIO (AI search optimisation)?
AIO is the work of making content readable and quotable by AI-powered answer engines such as ChatGPT, Perplexity and Google AI Overviews. It does not replace classic SEO; it is added on top of it.
Is an llms.txt file really necessary?
It is not a mandatory standard, but the cost is very low and the benefit is clear. By summarising in plain text what the site is and what its important pages cover, it makes it easier for AI readers to grasp the site correctly.
Does appearing in AI answers increase traffic?
Direct click numbers are lower than in classic search, but the quality of the traffic that does arrive is high and there is a clear rise in brand searches. Measurement should be done through referral sources and brand search volume.