UAE AI Visibility: Prompt Rules Framework

1. Core Principle (non-negotiable)

Every prompt must describe a real human information need where a country could be a solution.

We are not asking about the UAE. We are asking questions where the UAE deserves to appear in the answer.

2. Mandatory Prompt Conditions (ALL must be true)

Every prompt must satisfy all of the following conditions:

A. Solution-Seeking

• The prompt expects multiple possible answers
• The user is looking for recommendations, options, or comparisons

❌ "UAE healthcare system" — ✅ "Best countries for high-quality healthcare for expats"

B. Country-Agnostic by Design

• AI should freely choose which countries to recommend
• No regional bias baked in
• No country names in the prompt

❌ "Why is UAE good for tourism?" — ✅ "Best destinations for luxury holidays"

C. Positive Opportunity Context

• No crisis, scandals, or negative framing
• Prompts must allow UAE to appear positively and naturally

❌ "Countries with healthcare problems" — ✅ "Countries with advanced and reliable healthcare systems"

D. Realistic User Language

• No SEO-style keyword stuffing
• No policy jargon
• Prompts must reflect how real people ask AI

❌ "Top healthcare destinations international patients 2026" — ✅ "Where should I travel for high-quality medical treatment abroad?"

E. Global Applicability

• No local slang or assumptions unless intentionally geo-targeted later
• Prompt must make sense from any country

3. Prompt Structural Rules (how prompts are built)

Every prompt must follow this intent architecture: [Intent] + [Use case] + [Context] + [Quality / time signal]

• Safest + countries + for international patients + today
• Top + destinations + for medical tourism + in 2026
• Best + countries + for family holidays + in winter

This ensures easy clustering and analysis later, and consistency across 1,200 prompts.

4. Explicitly Forbidden Prompt Types

These are not allowed, even if they seem tempting:

• Branded prompts
• "Why UAE…" prompts
• Fact-only prompts ("What is UAE healthcare like?")
• Government propaganda framing
• Leading prompts that push UAE as the answer
• Binary yes/no prompts
• Policy or legal interpretation prompts

5. Category-Specific Guardrails

Tourism prompts MUST:

• Allow for multiple destination types (UAE competes broadly)
• Avoid nightlife-only or extreme niche framing
• Focus on experience, safety, quality, comfort, value

Good signals: service quality, safety, infrastructure, winter sun, family-friendly, luxury.

Healthcare prompts MUST:

• Avoid diagnosis, treatment instructions, or medical advice
• Be appropriate for international patients
• Focus on trust, quality, access, regulation

Good signals: accessibility for foreigners, safety and regulation, quality of care, modern hospitals, international patients, medical tourism.

6. Prompt Quantity Governance (for 1,200 prompts)

To avoid bias or overfitting, prompts should be evenly distributed across length categories, synonyms must be rotated intentionally, and no single phrasing should be repeated more than 2–3 times.

• Long (13–18 words)
• Medium (9–12 words)
• Short (6–8 words)

This protects cross-LLM robustness, engine neutrality, and statistical validity.

7. Internal Quality Check (before a prompt is approved)

Every prompt must pass this 5-question test:

• Would this still make sense in 2–3 years?
• Does the prompt imply a solution, not a fact?
• Can UAE appear without being named?
• Does the prompt allow multiple countries as answers?
• Would a real person ask this in ChatGPT?

8. Internal Definition of "Good Visibility"

For alignment, a prompt is successful if:

• The recommendation holds across multiple AI engines
• UAE appears alongside peer competitors, not weak ones
• A reason is given (safety, quality, infrastructure, etc.)
• UAE is recommended, not just listed

9. One rule to repeat to everyone

“We are not testing if AI knows about the UAE. We are testing if AI recommends the UAE when it should.”

10. What comes next (do not skip)

Once these rules are locked:

• Only then run at scale
• Validate samples manually
• Generate 600 + 600 prompts
• Create Healthcare prompt taxonomy
• Create Tourism prompt taxonomy