Generative engine optimisation (GEO) is the practice of making your brand easy for AI assistants such as ChatGPT, Gemini, Perplexity and Google AI Overviews to find, trust and cite in their answers. It combines technical access for AI crawlers, a clear and consistent brand entity, quotable content and third-party mentions. In 2026 it sits alongside SEO rather than replacing it.
Key takeaways
- AI engines choose sources differently: only about 11% of domains are cited by both ChatGPT and Perplexity, so one tactic won't cover every engine.
- Ranking first on Google no longer guarantees a citation. Only 17–54% of AI Overview citations come from the organic top 10.
- About 73% of sites have barriers that stop AI crawlers reading them. Fixing access is the fastest win.
- Most GEO happens off-site: Reddit is the most cited domain overall, and community and third-party sources make up nearly half of citations.
- llms.txt is optional; no major AI company has publicly committed to using it.
Buyers still search Google, but more and more they ask an AI assistant instead: "Which agency in Dubai is best for…" or "What's the best CRM for a UAE startup?". Google says AI Overviews now reach more than 2.5 billion people a month and AI Mode more than a billion, and AI Mode has been available in Arabic since late 2025. When an AI Overview appears, clicks to top organic results fall sharply, while the brands named inside the answer capture the attention. This guide explains how to become one of those brands.
What is GEO (generative engine optimisation)?
GEO is the set of technical, content and reputation practices that raise the chance an AI assistant mentions or links to your brand when answering a relevant question. Where SEO aims for a ranking position, GEO aims for inclusion in the generated answer, ideally with a citation link.
GEO isn't a trick or a single file. It builds on SEO foundations (crawlability, relevance, authority) and adds three things: content structured so a language model can lift a clean answer from it, an entity that models can identify without ambiguity, and a footprint on the third-party sites these models rely on. Some practitioners call it AEO (answer engine optimisation). The work is the same.
- Monthly users of Google AI Overviews (Google I/O 2026)
- 2.5B+
- CTR change for top-ranking pages when AI Overviews show (Ahrefs)
- −58%
- Domains cited by both ChatGPT and Perplexity
- 11%
- Sites with AI-crawler access barriers
- 73%
How do AI assistants choose which brands to mention?
Each engine uses different sources and habits. ChatGPT leans heavily on Wikipedia, reference media and Reddit, and mentions brands often without linking. Perplexity prefers fresh, specialist sources and forums. Google AI Overviews favour brand-owned domains most strongly and follow the organic results most closely.
| Engine | Source preference | Linking behaviour | What to prioritise |
|---|---|---|---|
| ChatGPT | Wikipedia (about 48% of top-10 citations in one study), reference media, Reddit | Mentions brands often, links rarely | Entity clarity, Wikipedia-grade facts, community mentions |
| Perplexity | Recent specialist sources, forums (about 17% of citations) | Balanced mentions and links, drives more clicks | Freshness, specialist depth, visible "updated" dates |
| Google AI Overviews / AI Mode | Brand domains (59.8% of citations) and pages close to the organic SERP | Most clickable links | Strong SEO, structured pages, answer-first sections |
| Gemini | Google index and knowledge graph | Varies by surface | Organization schema, Google Business Profile, consistent facts |
Figures from Otterly's AI Citations Report 2026 and Profound's platform citation patterns. Treat them as directional; engines change their behaviour often.
Why doesn't ranking #1 on Google guarantee an AI citation?
Because AI engines assemble answers from many sources, not just the top result. Since the May 2026 core update, only 17–54% of AI Overview citations come from organic top-10 results, and across engines only about 38% of AI-cited sources rank in Google's top 10.
The traffic maths has changed too. Ahrefs found AI Overviews correlate with a 58% lower click-through rate for top-ranking pages (The Next Web), and Pew found only 8% of users click a traditional result when an AI Overview appears, against 15% without one. A top ranking is still valuable, because it feeds AI Overviews, but it is no longer the whole game.
Step 1: How do you let AI crawlers in?
Allow the AI search crawlers in robots.txt, check that your CDN or firewall doesn't block them, and make sure key content is in the server-rendered HTML rather than loaded by JavaScript. About 73% of sites have at least one of these barriers (AuthorityTech).
Practical checks:
- robots.txt: explicitly allow the crawlers used for AI search and answers (OpenAI's search bot, PerplexityBot, Google's crawlers and others). Decide separately whether you allow training crawlers.
- CDN and bot protection: Cloudflare and similar services can block AI bots by default. Review the setting.
- Rendering: animated counters that render "0" in the HTML, or content that only appears after scripts run, is invisible to many bots. We saw this on competitor sites in our own market research.
- Speed and stability: fast, stable pages get crawled more completely. This is where web design and development and SEO overlap.
Step 2: How do you make your brand entity unambiguous?
Make sure every source that describes your business says the same thing: name, what you do, where you operate, who runs it and key facts. Then express those facts in structured data. Language models trust entities they can confirm across several independent sources.
- Organization and LocalBusiness schema on your site, with
sameAslinks to your LinkedIn, Google Business Profile, Clutch and other profiles. - One canonical boilerplate (under 50 words) used identically across your site, LinkedIn, directories and press.
- Author and Person schema for the experts behind your content.
- Consistent NAP (name, address, phone) across the UAE directories and maps.
Note that Google removed FAQ rich results on 7 May 2026 (Search Engine Journal). FAQPage markup is still valid schema and harmless, but don't expect a SERP feature from it. There is no AI-Mode-specific schema.
Step 3: How do you write content AI can quote?
Write reference-grade content: a clear definition near the top, question-style headings answered in the first 40–60 words, sourced statistics with dates, comparison tables and original data. Models lift clean, self-contained passages. Vague, padded paragraphs rarely get quoted.
Patterns that work:
- Definition blocks. "GEO is…" in one or two sentences.
- Answer-first sections. Each H2 answered directly, then expanded.
- Attributed numbers. "TikTok has the largest ad audience in the UAE, 12.5M adults (DataReportal, Digital 2026)."
- HTML tables for comparisons and prices, not images.
- Original data. A monthly index of which brands AI engines cite in your sector is the kind of asset journalists and models both reference.
- Local specificity. PDPL consent rules, Ramadan seasonality and Arabic search behaviour are details global content can't fake.
One vendor study reported 3–5 times more citations for reference-grade, well-chunked content. It's a correlation, not proof, but it matches what we see in practice. Google's 2026 core updates also rewarded first-hand experience and punished "made to rank" content, so quality serves both SEO and GEO.
Step 4: How do you earn third-party mentions?
Get your brand discussed on the sites AI engines read: review platforms, industry directories, Reddit and forums, news and trade media, and partner websites. Reddit is the most cited domain overall, and community and other third-party sources account for about 47.5% of citations.
- Reviews: Google Business Profile and Clutch reviews with specific detail.
- Directories and listicles: accurate profiles on the platforms buyers and models use.
- Community: helpful, transparent participation in relevant subreddits and forums (never astroturfing).
- Digital PR: original data and expert commentary in UAE and GCC media.
- Partners: case studies and listings on technology partner sites.
This is why GEO is closer to reputation building than to on-page SEO. Most of the work happens off your website.
Does llms.txt matter?
Not much, yet. About 10% of 300,000 domains have adopted llms.txt, Google has said it doesn't support it, and one log study found only 408 requests for it across more than 500 million AI bot visits over 90 days (Limy).
It is cheap to add and can help developer and agent tooling, so there's no harm in publishing one. But it won't fix blocked crawlers, weak content or a missing reputation. Spend your time on steps 1 to 4 first.
Why is GEO in Arabic an under-served opportunity?
Because demand exists and competition is thin. Google expanded AI Mode to Modern Standard Arabic in October 2025 across the Arab world (Google MENA), yet Arabic results for many commercial queries are dominated by small, low-authority sites.
A UAE or Saudi brand that publishes native, well-structured Arabic content, with consistent entity data in both languages and hreflang set up correctly, has a real chance to become the default citation for its category. Avoid scaled machine translation: Google's May 2026 core update hit it hard.
How do you measure AI visibility?
Track four signals: AI-assistant referral traffic in GA4, mention and citation rates across a fixed panel of buyer prompts, impressions-versus-clicks trends in Search Console, and self-reported attribution on your enquiry form. Together they show whether GEO is working.
| Method | What it tells you | How to set it up |
|---|---|---|
| GA4 "AI Assistants" channel group | Sessions and conversions from AI tools | Custom channel group matching chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com, claude.ai and others |
| Prompt tracking | Mention rate, citation rate, position, sentiment, competitor share | Around 50 buyer prompts in English and Arabic, checked monthly with a tool or scripted run |
| Search Console | Divergence between impressions and clicks | AI Mode and AI Overview traffic is folded into "Web"; watch informational pages |
| Enquiry form | The most honest signal of AI influence | Add "ChatGPT / AI assistant" to "How did you hear about us?" |
Also check what AI engines say about you, not just whether they mention you. In September 2026, some sources reported Google AI Overviews as "officially launched in the UAE" before that was true, a reminder that AI answers can be wrong and that clear, consistent facts protect your brand.
Our SEO and GEO agency team in Dubai runs this measurement monthly, and our marketing technology service sets up the GA4 channel group. For the budget side, see what SEO and GEO cost in Dubai.
FAQs
Is GEO replacing SEO?
No. GEO builds on SEO. Google AI Overviews still draw heavily on well-ranked, well-structured pages, and crawlability and authority matter for every engine. GEO adds entity work, quotable content and off-site reputation.
How long does GEO take to work?
Technical fixes such as unblocking crawlers can have an effect within weeks, as engines re-crawl. Entity and reputation work usually takes several months to shift mention rates, similar to SEO timelines.
Can I pay to appear in ChatGPT answers?
Not in the organic answer. ChatGPT Ads, available to UAE advertisers since September 2026, appear in a labelled sponsored box below the answer. Organic mentions still depend on GEO.
Do I need an llms.txt file?
It's optional. Adoption is low, Google doesn't use it, and major AI companies haven't committed to it. Add one if it's easy, but prioritise crawler access, content and reputation.
Should I create content in Arabic for GEO?
If your buyers use Arabic, yes. AI Mode supports Modern Standard Arabic and Arabic commercial results are thin, so native Arabic content can earn citations faster than English.
Want to know how often ChatGPT, Gemini and AI Overviews mention your brand today? Book a growth audit and we'll include an AI visibility check across your key buyer prompts.
