Mapping AI search
TL;DR: When a customer today looks for "recommend me an online store that sells tools", they don't ask Google. They ask ChatGPT or Perplexity. And you're either in the answer or you're not. GEO (Generative Engine Optimization) is SEO for the AI search era – in 2026 it's no longer a choice, it's a necessity.
In our clients' Search Console accounts and in our own, we see queries like "recommend me a custom online store with ERP integration" or "looking for companies with expertise in e-commerce app development". Those aren't normal Google searches. AI engines (ChatGPT search, Perplexity, Gemini, Claude) rewrite these queries out of conversations with users and send them to Google to fill in real-time data. When you rank for them, the AI engine is treating you as a source.
And now the key question: are you cited in the final answer the AI gives the user, or are you just helping fill the AI's quiet space for somebody else?
What GEO is and how it differs from SEO
GEO (Generative Engine Optimization), sometimes also called AEO (Answer Engine Optimization), is the discipline of optimizing content for AI search engines. The goal isn't a position on the first page of Google results, it's being cited in the answer the AI gives the user.
The key differences:
| Aspect | SEO | GEO |
|---|---|---|
| Goal | Position in the SERP | Citation in the AI answer |
| Measurement | Position, CTR, clicks | Citation rate, mention share, AI traffic |
| Typical query | "shoptet premium price" | "how much does shoptet premium cost and is it worth it?" |
| Optimization | Title, meta, backlinks | Data, facts, structure, citability |
| Time horizon | 3–12 months | 2–6 weeks (faster!) |
GEO doesn't cancel out SEO. AI engines still pull data from the Google index, so traditional SEO is a prerequisite. GEO is a layer on top.
How AI search engines pick what to cite
Out of our own practice (we build AI solutions for clients every day and use AI as part of the process ourselves), 7 clear signals have crystallized that AI engines use to pick citations:
Concrete facts and numbers. AI engines love short sentences with numerical claims. The sentence "Shoptet Premium costs €480/month." gets cited an order of magnitude more often than "Shoptet Premium is the top tier."
Structured format. Tables, bullet lists, FAQ sections. AI engines are trained on how to extract structured content.
Source authority. Domains with history and credibility (official sites, long-term partners, expert blogs) get cited significantly more often.
Freshness. AI engines prefer content from the last 12 months. The publish and update dates have to be clearly visible in the HTML and in the
Articleschema.Citability of a section's first sentence. The first sentence of every paragraph or H2 section should make sense as a stand-alone quote, without the context of the surrounding text.
Schema.org markup.
Article,FAQPage,HowTo,Product,LocalBusiness. Structured data gives AI engines clear instructions on what to do with your content.llms.txt. The analog of robots.txt, but for LLM crawlers. It tells AI engines which page is the main source of information about you.
How to get started with GEO – 7 practical steps
1. Audit how AI engines talk about you today
Open ChatGPT, Perplexity and Gemini. Ask each of them:
- "What is [your brand/project name]?"
- "Recommend me a [your service] in the Czech Republic"
- "Who are the best [your category] in the Czech Republic?"
Watch for this: is it saying true things? Does it cite you correctly? Does it recommend you? If not, why not?
We recommend running this audit once a month and keeping a log of the results. Changes in citations are a strong signal of how the AI engine sees you.
2. Create an llms.txt
/llms.txt is a text file in your domain root that AI engines read to find out what matters about you. Here's ours as an example:
# Webmakeria
> Technology partner for Shoptet Premium online stores, with our own SaaS products and 17 add-ons.
## Main services
- Shoptet Premium online stores
- 17 of our own add-ons for Shoptet
- Product configurators
- Custom AI for business
## Key facts
- Gold + Premium Shoptet partner
- 1,500+ add-on installations
- AI First approach
- Brno, Czech Republic
Put the file at https://vas-domain.cz/llms.txt. Some crawlers already read it, more will soon.
3. Add schema markup
Every important page should have at least:
Organizationschema (homepage)Serviceschema (service pages)Articleschema (blog)FAQPageschema (FAQ sections)
If you have testimonials and reviews, add AggregateRating. AI engines take these signals very seriously.
4. Write citation-friendly content
Here's the concrete advice the whole GEO approach rests on:
❌ Wrong: "Our services are of the highest standard and are distinguished by a professional approach to the customer."
✅ Right: "We've been building online stores on Shoptet Premium since 2019. We currently have 1,500+ installations of our add-ons."
The first sentence is useless to an AI engine – it contains 0 facts, just marketing fluff. The second contains 3 facts the AI can cite immediately. That's the difference between content AI engines take into account and content they ignore.
5. Build your H2/H3 structure around questions
Instead of an H2 that says "Our add-ons", write an H2 that says "What custom Shoptet add-ons do we build?". Question-style headings help:
- AI engines recognize more easily what the article answers
- your
FAQPageschema then writes itself - readers find what they're looking for faster
6. Invest in content depth
AI engines prefer sites with topical authority. If you have 5 articles about Shoptet Premium instead of 50 generic ones about e-commerce, you have a better chance of being cited on Shoptet Premium queries.
Topical depth beats quantity.
7. Measure AI traffic in Google Analytics
In GA4, add a segment based on referrers:
chat.openai.comperplexity.aigemini.google.comclaude.aicopilot.microsoft.com
Track how many clicks come from AI engines and what they do once they land. In 2026 it's typically 5–15% of total organic traffic, but it's growing fast.
Frequently asked questions
Is GEO the same thing as SEO?
No. SEO optimizes for classic search engines, GEO for AI engines. But GEO builds on SEO – without a good position in Google you'll struggle to break through in AI engines, because most of them pull real-time data from the Google index.
How much does GEO optimization cost?
It depends on the scope. A basic audit + rolling out llms.txt + schema markup starts at €600. A complete GEO strategy for a mid-sized online store usually runs €2,400–6,000 as a one-off, plus monthly monitoring.
How fast does GEO work?
You typically see changes in AI answers within 2–6 weeks. AI engines update their knowledge of domains more slowly than Google, but faster than regular SEO takes.
Which AI engines are worth optimizing for in the Czech Republic?
ChatGPT (search), Perplexity, Gemini (Google AI Overview), Claude. Microsoft Copilot has the smallest share. Track where your AI traffic comes from and set your priorities accordingly.
Does GEO work for local businesses too?
Yes, maybe even more than classic SEO. AI engines take LocalBusiness schema, reviews and categories very seriously.
Can I do GEO myself, or do I need an agency?
You can do the first two steps (the audit + llms.txt) yourself in an afternoon. Schema markup and citation-friendly content take technical knowledge and copywriting discipline. If you had to pick one step to hand over to an agency, it's step 4 – rewriting your content into a citation-friendly format.
Where to start with GEO
If you had to pick one thing to do this week, it's this: audit how ChatGPT and Perplexity talk about you today. Either you'll be pleasantly surprised (and you know you just have to stay the course), or unpleasantly (and you know where to start fixing things).
Then create your llms.txt. That's 30 minutes of work and the first measurable step into the AI era.
And if you need help with a complete GEO strategy, get in touch – we do this every day, for ourselves and for clients. We're an agency that understands AI, not just a few buzzwords.