Measurement protocol / revision 1

What the numbers
actually mean.

Pathmeter reports what your browser can move over one HTTPS connection to this node. It does not estimate your tariff or the theoretical capacity of the access link.

Phase A

Warm the route

The browser makes five uncached requests to the node. The median round-trip duration is shown as latency, reducing the effect of a single scheduling spike.

5 samples → median RTT
Phase B

Stream the payload

The chosen test object is read as a browser stream. Bytes are counted as they arrive, without waiting for the entire payload to fit in memory.

bytes × 8 ÷ elapsed time
Phase C

Report the run

The graph follows the cumulative transfer rate. The final value describes the complete run; stopping early leaves the partial measurement visible.

application throughput ≠ line rate
Included

What affects the reading

Every layer between the browser and the node participates: local Wi-Fi, access-network congestion, provider routing, interconnection capacity, packet loss, TCP congestion control, TLS processing and browser scheduling.

That is intentional. Pathmeter measures the usable application path, because that is the path a real download uses.

Not a multi-flow testOne transfer avoids hiding a weak route behind many parallel connections.
Not a latency laboratoryBrowser timing includes normal client and server scheduling overhead.
Not a permanent scoreResults belong to one route, destination and moment in time.