How we rank
Every score and ranking on Healthy: Find & Compare Local Medical Centers comes from the same published rules, applied to every business. This page explains the data we use, how it becomes a score, and where its limits are.
Principles
- Evidence over averages. We rank on what reviewers describe, and we show the evidence: how many reviews mention a thing and how positively.
- Same rules for everyone. No business can pay for placement, and nothing is adjusted by hand for individual businesses.
- Small samples count less. A handful of glowing reviews should not beat years of consistent ones, so every figure is adjusted for how much evidence stands behind it.
- Missing is not negative. When we lack data on a topic, we leave it out. We never assume a business lacks something just because nobody mentioned it.
The data we use
- Public listing details: name, address, category, opening hours, price range and the features a business lists for itself.
- Public ratings: the overall star rating, the number of reviews and how they split across one to five stars.
- Review text: a large sample of individual public reviews per business, including the structured answers some reviewers give (for example a separate rating for communication, staff friendliness, professionalism, and wait time), the date of each review and whether the owner replied.
We do not use paid placements, private data, or anything a business sends us privately.
Reading the reviews
Software reads each review on its own and records what it is about: which treatments it mentions and whether it speaks well of them, how it rates the communication, staff friendliness, professionalism, and wait time, what the visit cost, how long the wait was, the occasion, and any specific problems it reports. The output is checked against a fixed format before it is stored, and anything that does not fit is discarded. Reviews written in other languages are read in translation.
Our score (0–10)
Our score is not a star rating. It combines up to three parts, each scaled to 0–1:
| Part | Weight | What it measures |
|---|---|---|
| Rating quality | 50% | The public star rating, adjusted for the number of reviews behind it. |
| Experience | 30% | The average of what reviewers rated for communication, staff friendliness, professionalism, and wait time, with recent reviews counting more. |
| Praise vs. complaints | 20% | The share of positive themes (friendly staff, value, cleanliness…) among all themes reviewers raise. |
When a part has no data for a business, the other parts are re-weighted, so a business is never penalized for missing information. The result is multiplied by 10 and rounded to one decimal.
Adjusting for the number of reviews
Each rating is blended with the average rating in its category, as if it came with 20 extra reviews at that average. With a category average of 4.3, a perfect 5.0 from 4 reviews counts as 4.42, while a 4.7 from 2,000 reviews stays at 4.70. As reviews pile up, the adjustment fades and the business's own record takes over.
Recent reviews count more
In the experience part, a review's weight halves every 365 days, so a review from a year ago counts half as much as one written today. A business that recently improved, or slipped, shows it there first.
Rankings for specific treatments
"Best" lists rank businesses on one thing, such as a dish or a service, within a city or neighborhood.
- Qualifying: a business needs at least 3 positive mentions of the item to be ranked for it. A ranking is published only when at least 6 businesses in the area qualify.
- Order: 65% how positively reviewers describe the item (adjusted for the number of mentions, like ratings above), 20% the business's overall rating quality, and 15% how often the item comes up, on a scale that flattens out so sheer volume cannot dominate.
- Freshness: closed businesses, and businesses whose data is more than 180 days old, are left out of rankings.
Features and filters
Feature pages (wheelchair access, evening hours and so on) combine what a business lists for itself with what reviewers report. Each feature resolves to yes, no or mixed, and the page shows how many signals support it. A filter page is published only when at least 8 businesses in a city have the feature.
"Skip if" warnings
A problem appears as a warning only when it is a pattern: at least 2 analyzed reviews must report it, and they must make up at least 25% of the business's analyzed reviews. One angry review is not enough. Warnings name a specific problem (for example a hard-to-find entrance or long waits for a callback), not a general mood.
Which businesses get a page
- We cover businesses with at least 10 public reviews.
- A business page is published for search engines only with at least 4 analyzed reviews and at least 4 sections backed by real data. Thinner pages stay available to visitors but are kept out of search results.
- Businesses listed as permanently closed keep their page with a clear notice, are not shown in search results, and are removed from rankings.
- Businesses outside the categories this site covers are left out, even when they show up in the source data.
Limits
- Reviews come from people who chose to write one, so they lean toward strong opinions.
- Some reviews on any platform are fake or incentivized. Adjusting for sample size and requiring patterns before warnings limits their effect, but cannot remove it.
- We analyze a sample of reviews for each business, not necessarily every review ever written.
- Listings can be out of date. Each page shows when its analysis was last updated; check with the business before you go.
Found a mistake? Email reviews@us.healthy.tn with the page address and what is wrong. We correct factual errors, but we do not change scores or remove criticism on request.