Methodology

TestConn measures against Cloudflare’s global edge network rather than servers we operate. Your browser exchanges data directly with the Cloudflare location closest to you, which keeps the measurement representative of real internet performance.

Because the measurement traffic goes to Cloudflare, Cloudflare necessarily receives that connection — including your IP address — in order to perform the test. Where the measurement library supports it, TestConn disables the additional results logging that would otherwise send computed scores onward.

Your browser downloads and uploads sample payloads of increasing size and times each transfer. Starting small and ramping up lets the connection reach full throughput before the measurement settles, which is why the live number climbs during the test. Results are reported in megabits per second (Mbps).

Ping (latency) is measured from many small round-trips to the edge, reported in milliseconds. Jitter is how much those round-trip times vary between one another — a steady connection has low jitter even if its average ping is not especially fast.

Measuring packet loss reliably from a browser requires an additional server component that TestConn does not currently operate. Rather than display an estimate that could mislead you, packet loss is reported as N/A until it can be measured properly.

Your public IP address is determined server-side from the incoming request. It is then used for a country and network-operator lookup with a third-party IP data provider, which returns the provider name, ASN and country. This lookup happens on our server so that no API credentials are exposed in your browser.

Location is approximate and country-level only. TestConn does not determine a city, street or precise position, and cannot reliably detect VPN or proxy use — so if you are on a VPN, the details shown belong to that service rather than your home provider. See IP & Network Information.

A browser-based test measures what your browser could achieve at that moment, on that device, over that connection. Wi-Fi signal, distance from the router, other devices, background downloads, VPNs, congestion and your device’s own limits all affect the outcome.

Because of this, your result may differ from other testing services — they use different servers, payload sizes and methods. Treat TestConn as a consistent gauge over time rather than an absolute verdict, and test over Ethernet with other activity paused for the most repeatable reading.

Measurement results and network details are shown to you in your browser and are not stored by TestConn. Full IP addresses are not retained or written to logs or analytics. The Privacy Policy sets out the detail, including the third parties involved.