What ChatGPT recommends for Riyadh clinics
A worked, anonymised look at which kind of clinic ChatGPT names when asked for a recommendation in Riyadh, and the traits the named clinics share.
Asked plainly for a good clinic in a Riyadh district, ChatGPT tends to name a small, repeating group of practices, and those practices usually share three traits: complete listings, a steady flow of recent reviews, and pages that state plainly what they treat, where, and for how much.
That pattern is one worked example inside the larger question of how Saudis use AI search to find businesses; this post walks through it in detail, using a hypothetical clinic to show the method without naming anyone real.
On this page
What did we actually ask, and why does that matter?
We asked a small set of plain, buyer-style questions such as "which dental clinic in north Riyadh is good with kids and takes Bupa?" and "is [clinic name] any good?" — the phrasing a real patient would use, not a keyword like "dentist Riyadh". The wording matters because an assistant answering a full question pulls from different pages than one answering a bare keyword; it is looking for a passage that answers the specific condition, not just the topic.
Running the same questions in Arabic and in English is part of the method too, because the two languages frequently return different names. We cover that comparison on its own in Arabic vs English prompts.
We also ran each question more than once, on different days, before drawing any conclusion. A single conversation with an assistant is a sample of one, and the sources behind an answer can shift from run to run even for the same wording. Treating one lucky or unlucky answer as the truth is the most common mistake we see clinics make when they test this themselves, usually after a single bad result sent to them on WhatsApp by a worried colleague.
What did the named clinics have in common, in this worked example?
In our hypothetical run, three clinics came up repeatedly across the district-level questions. Naming them here would misrepresent them as a real result, so we describe the pattern instead of the names.
A complete, verified Google Business Profile
Full hours, phone number, accepted insurance and a description written in plain language, not a slogan. Google's own guidance treats a claimed and verified profile as the baseline for a business to be found and trusted at all [2].
A steady, recent flow of reviews
Not the highest star rating in the district, but the most consistent one — reviews from the last few months, not a cluster from three years ago.
Pages that state facts, not feelings
"We treat braces and clear aligners for adults and teens, open until 11pm, in Al Olaya" rather than "Your smile, our passion." The factual version is easier for an assistant to lift and quote.
Structured data behind the facts
LocalBusiness markup that matches what the page says in plain text, so the two signals confirm each other instead of contradicting [3].
What did the clinics that were never named have in common?
The opposite pattern, mostly. In the hypothetical set we tested against, the clinics that never appeared tended to have an outdated or unclaimed Business Profile, a review count that had stalled for a year or more, or a website built entirely around a hero image and a phone number with no real description of services underneath it.
| Trait | Tended to be named | Tended not to be named |
|---|---|---|
| Business Profile | Complete, claimed, current hours | Unclaimed or years out of date |
| Reviews | Steady flow, recent dates | Old cluster, then silence |
| Website copy | States what is treated, where, for whom | Slogans with no service detail |
| Crawler access | Assistant search crawlers allowed | Blocked by a default security rule |
That last row is easy to miss. OpenAI documents which of its crawlers fetch pages during a live conversation versus training, and a clinic's own hosting provider or firewall can block the wrong one without anyone noticing, silently removing the clinic from answers it should otherwise have earned [1].
What would we actually recommend a clinic do first?
Start with access, then facts, then trust. Confirm the crawlers can reach your site, claim and complete your Google Business Profile in both Arabic and English, then rewrite your service pages so the first sentence under each heading states plainly what you treat and for whom.
Reviews come last in the fixing order but first in the buyer's mind, so keep asking happy patients for one every week rather than in occasional bursts. A clinic that does the first three things well tends to see the fourth follow naturally, because a good, well-documented clinic is simply easier for real patients to review too.
One more detail worth stating plainly: none of this is about writing content aimed at the assistant instead of the patient. Every one of the four traits above — a complete profile, honest reviews, plain factual pages, working crawler access — is also just good practice for a real patient trying to choose a clinic on a normal day. The assistant is reading the same signals a careful human would, which is the whole reason this kind of work holds up over time instead of being a trick that stops working the next time a model updates.
How do you know if it worked, without guessing?
Re-run the same fixed set of questions every month and record two things: whether you were named, and whether a competitor was named instead. A single good or bad answer on one day means little, because the exact sources an assistant draws from shift from run to run. A trend across a quarter means something.
If the pattern in this worked example matches what you suspect is happening to your own clinic, the wider guide on AI adoption in Saudi Arabia explained covers why this behaviour is only going to become more common, and AI Overviews in Arabic covers the parallel version of this same test on Google's own AI features.
Questions people also ask
Does ChatGPT search the live web when it recommends a clinic?
It can, using a search feature that fetches current pages rather than relying only on older training data. That is why a clinic's current pages and listings matter, not just whatever existed when a model was trained.
Is this a real client result?
No. RankPearl has no clinic clients we can name here, so this is a hypothetical, clearly labelled worked example built from the pattern we see when we run this test for real prospects, not a case study of an actual account.
Can a clinic guarantee it gets recommended?
No, and be wary of anyone who says otherwise. Nobody controls an assistant's output. A clinic can remove the things that stop it being read, and improve the odds, which is different from a guarantee.
How often should a clinic re-run this kind of test?
Monthly, with the same fixed set of questions. A single run tells you today's snapshot; a monthly run tells you whether your changes are working.
Sources
- Overview of OpenAI crawlers — OpenAI (checked 23 September 2026)
- Add or claim your Business Profile — Google Business Profile Help (checked 23 September 2026)
- LocalBusiness structured data — Google Search Central (checked 23 September 2026)
Run the same prompt test on your own clinic
Book a free 30-minute consultation with Ibrahem and Jeremy. We look at your site while we talk, and you leave with a prioritised plan.
Who wrote this
Ibrahem HelalCo-founder
Leads strategy and delivery.
Works on: SEO, GEO
Reviewed before publishing by Jeremy Zava, Co-founder.
- How Saudis use AI search to find businessesHow people in Saudi Arabia use ChatGPT, Gemini and AI Overviews to pick businesses, and what a Saudi company must fix to be the name they get back.
- AI adoption in Saudi Arabia explainedHow many people in Saudi Arabia are online, and how that connectivity turns into ChatGPT, Gemini and AI Overviews use for everyday buying decisions.
- AI Overviews in Arabic search resultsHow Google's AI Overviews behave on Arabic search results in Saudi Arabia, where they differ from English ones, and what it means for your Arabic pages.
- Arabic vs English AI prompts: what changesWe compared how ChatGPT, Gemini and AI Overviews answer the same buying question in Arabic and in English, and what the gap means for your content.