Test methodology

What we measure

The test measures the path between your device and the nearest Cloudflare data center (colo, short for "colocation") — typically Madrid or Barcelona for users in Spain. We measure:

• Download and upload: we transfer incompressible data blocks of increasing size and report a percentile-based summary of observed throughput, not a simple average.
• Idle latency: around 20 small requests before the connection is loaded, to get baseline latency and jitter (variation).
• Loaded latency: the same measurement while the connection is saturated with download/upload traffic — this reveals "bufferbloat" (how much your connection degrades while busy), a signal most Spanish speed tests don't show.

Why we show the colo

Every result includes the Cloudflare colo that served it (for example MAD for Madrid). It's the correct, consistent yardstick for comparing carriers within Spain, but it means the test measures your connection up to that point in Cloudflare's network, not to an arbitrary destination on the internet.

Why results can differ from Ookla or other tests

Every speed test measures against its own infrastructure with its own ramp-up methodology. Differences in the measurement point, the negotiated network protocol (HTTP/2 vs HTTP/3), and your network conditions at that moment explain normal variation between tests.

Honest limitations

Results are indicative, not a contractual measurement of your carrier's service — in Spain, contractual speed disputes use the government-certified test. Wi-Fi, VPN use, and your device's own limits all affect the result. The test automatically pauses if you switch tabs mid-measurement, so it doesn't record a false result.