We publish only what we can substantiate. This page documents precisely how our tool measures a connection, how we source country- and provider-level comparisons, and how we protect your privacy. If you cite our data, this is the reference to link.
Every test returns three core metrics:
SpeedCheckTest runs entirely in the browser — no app, no account, no install. A test has three phases: a ping phase built from repeated lightweight HTTP round-trips (we report the minimum, which best reflects line latency), a download phase using multiple parallel streams, and an upload phase using parallel requests. The tool auto-selects the nearest of six global server locations to reduce routing noise. This approach follows the transparent, open-source LibreSpeed methodology used by many reputable tools.
Because a browser test measures your connection at that moment, results vary with Wi-Fi signal, other devices, the test server, and time of day. For a number you can act on, we recommend the setup in how to test internet speed accurately: wired connection, background apps closed, and multiple runs.
Our country speed index and rankings content is grounded in independent, publicly documented benchmarks rather than single-tool claims. The primary sources are:
When a visitor runs a test and explicitly accepts the consent banner, one anonymized result may be recorded: ping, download, upload, the test server, language, and a coarse region derived from Cloudflare's network headers. The raw IP address is never read or stored. Visitors who decline are excluded entirely. Aggregates are published only above a minimum sample size so small or noisy samples never drive claims. Full details are in our privacy policy.
We provide methodology walkthroughs, aggregate data, and chart-ready assets on request. See the press & media kit or contact [email protected].
Yes — the tool runs a live, multi-phase browser test (ping, download, upload) against the nearest of six global servers. Because it runs in the browser, results reflect your environment at that moment; use a wired connection and repeat runs for the most representative number.
From independent, publicly documented benchmarks — the Ookla Global Index, the ITU, the FCC's Measuring Broadband America, national regulators, and firms like OpenSignal. Each comparison page cites the specific sources it uses.
No. Any recorded result is anonymized and consented; coarse region comes from Cloudflare's edge headers and the raw IP is never read or stored.
Medians resist outliers, so a handful of very fast or very slow results can't distort the typical figure. It's the standard for honest speed reporting.