We sell measurement discipline, so we apply it to ourselves first. This page is the method; the live scoreboard follows as soon as the first measured citations land — with screenshots.
1. A locked question basket. Ten questions a real buyer would ask about services like ours — branded, problem-intent and category questions, in English and Arabic. Locked before measurement starts; any change restarts the baseline.
2. The engines patients and buyers actually use. ChatGPT (free tier, temporary chat) and Gemini (Flash) measured automatically; Perplexity, Google AI Overviews and Grok in weekly manual windows. Never paid tiers — they are not what customers see.
3. Multi-run windows, not screenshots. A "measurement" is each question, run on each engine, daily across a window — because single runs are anecdotes. AI answers change day to day; our own data across Dubai clinics shows recommendations flipping within 24 hours.
4. Evidence on file. Every run is screenshotted and logged with date, engine and model tier.
5. Published either way. The scoreboard shows the climb at whatever speed it happens. No projections, no "typical results" — dated measurements only.
Anyone can show you one lucky screenshot. The question that decides whether AI visibility is real is: named how often, on which engines, measured the same way every week? That is the number we report to clients — and the one we hold ourselves to on this page.