Six criteria. Three directories. One of them loses two rows.
This desk reviews directory software. It scores every property on the same six criteria, publishes the criteria before the scores, and measures every number on the live site rather than repeating the press copy. The three under the knife this quarter are all built by labs.llc: two open registers of practices, and one live relay into a federal register.
Nothing here is medical advice, and none of these three products gives any. They are directories and lookups. Where they refuse to answer something, that refusal is scored as a strength.
The standing scoreboard
| Criterion | Dentist | RX | Clinicians |
|---|---|---|---|
| Reach | 4 | 5 | 1 |
| Currency | 3 | 3 | 5 |
| Candour | 5 | 5 | 4 |
| Addressability | 4 | 4 | 0 |
| Cost of an answer | 4 | 4 | 3 |
| Self-agreement | 3 | 3 | 4 |
| Total | 23 / 30 | 24 / 30 | 17 / 30 |
The two registers separate by a single point, and they separate on size alone. The relay loses two rows outright, and it loses them for one reason with two faces: it knows one country, and nothing it finds has an address.
The verdicts
-
23 / 30
Dentist — verdict: candid
109,030 practices in 69 countries, every city page printing its own gaps in the masthead — and a front door that opens on the one region the register has no addresses for.
-
24 / 30
RX — verdict: deep
The same engine over a heavier body: 141,701 doctors, Germany-led by more than two to one, 4,201 crawlable pages — and two different city counts on one screen.
-
17 / 30
Clinicians — verdict: narrow
A live federal register with a phone number on every row, asked for Berlin and answering with 135 Americans. The freshest of the three and the only one you cannot link to.
The criteria, published before the scores
Six criteria, nought to five each. They are the same six for every property this desk reviews, they do not move between reviews, and each one is a question that can be answered with a measurement rather than a mood. A directory that scores badly here is not a bad idea — it is a good idea with a specific hole in it, and the hole is named.
- Reach — 0 to 5
- How much of the world is actually in the record, and how much of it you are allowed to ask about. Counted as records and countries, not as ambition.
- Currency — 0 to 5
- How old the record is, and whether the page tells you without being asked. A dated snapshot beats an undated one; a live read beats both.
- Candour — 0 to 5
- Whether the property publishes its own gaps where a visitor will see them, rather than in a methodology note nobody opens. Completeness percentages printed in the masthead score here. So does a clear refusal.
- Addressability — 0 to 5
- Whether an answer has a URL a stranger can open. A directory whose results live only inside a session is a tool, not a reference, and cannot be cited, linked or indexed.
- Cost of an answer — 0 to 5
- The bytes and the seconds between the request and the first true number on the screen. Measured with the network, not estimated from the source.
- Self-agreement — 0 to 5
- Whether the property's own numbers agree with each other. Two counts of the same thing on one screen is the cheapest possible way to lose a reader's trust, and it is the one this desk checks first.
How these were measured
Every figure in these three reviews came from a plain public request to the live site on 22 September 2026, paced two to four seconds apart, and every one carries the UTC minute it was taken. Screenshots were captured with a headless browser at 1440 px with analytics endpoints blocked at the network layer, so nothing in the illustrations is a live tracking hit. Where a figure appears only after a control is pressed, the review says which control. Where the build tree and the live site disagreed, the live site won.
No rating markup is emitted anywhere on this site. The scores are visible numbers in an HTML table, meant for you, not for a search result.