Proxy Speed Test
Speed testing should answer a simple question: How fast is this route for this user under these conditions?
Measure more than one value
Record:
- Download speed
- Upload speed
- Test duration
- Test time
- Node and region
- Whether the connection is idle or already carrying traffic
Use repeated tests
One speed test can be an outlier. A better method is to run several tests using the same node, then look at the typical result and the spread between runs.
Test busy periods
Repeat the test at a normal period and at a busy period. Large differences can reveal congestion that a daytime-only test would miss.
Do not compare unrelated conditions
A Wi-Fi test on one ISP should not be treated as directly comparable to a wired test on another ISP. Record the test environment so that differences are understood before drawing conclusions.
Public measurement context
Cloudflare Radar publishes aggregated Internet-quality measurements, including estimated download speed, latency and DNS response time. M-Lab provides open measurement infrastructure and datasets that can help provide broader network context. These resources are not a substitute for provider-specific testing.
Recommended report
A useful test report can show the median download speed, median upload speed, range, time of day and the number of successful runs. This is more informative than publishing the fastest single number.