A probe placed on a site measures ping, jitter, packet loss, radio state and outages, around the clock, and runs a throughput test at a regular interval. You compare operators on dated figures, not on a coverage map. Each line has a data cap that is enforced before anything is consumed.
No credit card to start. Secure payment through Stripe when you change plan.
The axes, the cadences, the outage log and the 2 GB per line cap are real. The curve is not: it shows the shape of a series, and it will stay dashed until the first real run. We would rather show an empty, dated chart than one filled with an average bought elsewhere.
A product that hides its limits wastes everybody's time. Here are ours, today.
Ping, jitter and loss every 5 minutes, radio state every minute, throughput every hour. A one-off test tells you what happened at 2 pm; a probe tells you what happens at night, at peak hours, and during an outage.
Each line is capped at 2 GB per month. Before every throughput test the estimated usage is computed, and the run is refused if the budget would be exceeded: nothing is consumed, and the refusal is recorded with its reason.
Hour, day and week windows, average, min, max and p95, peak hours (6 pm–10 pm) against night (0–6 am) in local time. An operator can look good on average and bad at the hour when your lines actually work.
Every network loss is timestamped with its start, end and duration. Availability is only computed when an outage log exists: without one it is null, never a default 100%.
Figures come out through an API and a CSV export. A test compares both cell by cell: what you download is exactly what the API returns.
An operator holding 12 ms average latency can collapse at 6 pm. That is exactly what this grid must show, cell by cell, with dated figures. Today it is empty: the fleet holds one phone and one line, so four networks cannot be compared.
| Orange | SFR | Bouygues Telecom | Free Mobile | |
|---|---|---|---|---|
| Average latency (ms) | no measurement | no measurement | no measurement | no measurement |
| Jitter (ms) | no measurement | no measurement | no measurement | no measurement |
| Packet loss (%) | no measurement | no measurement | no measurement | no measurement |
| Outages (log) | no measurement | no measurement | no measurement | no measurement |
| Download throughput (Mb/s) | no measurement | no measurement | no measurement | no measurement |
The product can, but no comparison has ever been produced: the fleet holds a single device with a single SIM, all in one place. Until probes are placed at customer sites, we publish no operator figures.
Not yet: the speed-test server does not exist server-side, and no second site is deployed. The throughput columns stay empty rather than filled with an invented number, and a manual throughput run is refused without consuming anything.
No. One phone measures one place. National maps already exist and are not our job: we measure real quality, continuously, at one precise location that matters to a customer — and placing a probe at their site is what makes a multi-site comparison sellable.
A probe is a phone with a SIM and a charger, 60 to 120 € of hardware. Throughput costs data: hence the 2 GB per month and per line cap, and less frequent throughput tests on shared lines.
Create an account, take a key, run your first test. It is free to start.