Scoring Methodology
We believe clinic directories should be transparent about how they rank and score clinics. This page explains exactly how our system works — no black boxes.
1. Data collection
We collect publicly available Google Maps reviews for every verified medical clinic in Bangkok. This includes the review text, star rating, reviewer name, review language, and publication date. We also collect clinic metadata: name, address, phone, website, opening hours, and amenities.
Reviews are collected periodically. The “data collected” timestamp on each clinic profile shows when data was last pulled.
2. AI classification
Each review with English text (or an English translation) is sent to a large language model for classification. The model analyzes the review across 8 dimensions and extracts:
- Which dimensions are mentioned
- Sentiment per dimension (+1 positive, 0 neutral, -1 negative)
- Evidence phrases from the review text
- Whether the reviewer is likely a foreigner
- Specific medical topics mentioned
3. The 8 dimensions
| Dimension | What it measures |
|---|---|
| English Communication | Staff English fluency, clear explanations, translation availability |
| Wait Time & Efficiency | Appointment punctuality, queue length, administrative speed |
| Clinical Competence | Diagnosis accuracy, treatment outcomes, doctor knowledge |
| Pricing Transparency | Upfront quotes, surprise charges, value versus expectations |
| Facility Quality | Cleanliness, modern equipment, comfort |
| Staff Friendliness | Warmth, patience, willingness to help, empathy |
| Foreigner Accessibility | International insurance, location convenience, visa documentation, cultural sensitivity |
| Follow-up Care | Post-visit contact, aftercare instructions, continuity of care |
4. How scores are calculated
For each dimension, we aggregate all sentiment values from classified reviews:
score = round((raw_score + 1) / 2 × 100)
This produces a score from 0 to 100, where 100 means every mention was positive and 0 means every mention was negative. A score of 50 means an equal mix.
5. Confidence levels
We only display scores when we have enough data. Each dimension score has a confidence level based on the number of mentions:
| Mentions | Confidence | Display |
|---|---|---|
| 10+ | High | Score shown normally |
| 5–9 | Medium | Score shown normally |
| 3–4 | Low | Score shown with “low confidence” label |
| 0–2 | Insufficient | Score hidden |
6. Badge thresholds
Badges are awarded to clinics that consistently score well on specific dimensions. Each badge has a minimum score and minimum number of mentions:
| Badge | Dimension | Min score | Min mentions |
|---|---|---|---|
| Excellent English Communication | Communication | 75 | 5 |
| Transparent Pricing | Pricing | 70 | 4 |
| Foreigner Friendly | Foreigner Access ≥70 + Communication ≥65 | — | 5 combined |
| Quick & Efficient | Wait Time | 75 | 5 |
| Exceptional Staff | Staff Friendliness | 80 | 5 |
| Strong Clinical Reputation | Clinical Competence | 75 | 5 |
| Modern Facility | Facility Quality | 75 | 4 |
| Great Follow-up Care | Follow-up Care | 70 | 3 |
7. Warning flags
Clinics that score below 35 on any dimension with at least 5 mentions receive a warning flag. Warning text is displayed on the clinic profile page so patients can make informed decisions.
8. Limitations
- Review bias — Google Maps reviews skew toward extreme experiences. Satisfied patients who had an ordinary visit rarely leave reviews.
- Language coverage — We primarily analyze English-language reviews or translations. Clinics with mostly Thai reviews may have less dimension data.
- AI classification errors — Large language models can misclassify sentiment or dimensions. We mitigate this with volume: a single misclassified review has minimal impact when aggregated across many reviews.
- Point-in-time data— Scores reflect reviews collected up to the “data collected” date. Clinics can change between data refreshes.
- Not medical advice — This directory helps narrow your search. It is not a substitute for professional medical judgment.
Questions?
If you have questions about our methodology or believe a score is incorrect, contact us at hello@medicalclinicbangkok.com.