Match the conditions before comparing the numbers
Use the same device, connection type, location and test service. Record whether the connection is busy or a VPN is active. Run tests one after another, not simultaneously: simultaneous tests compete for capacity. A result from a phone over Wi-Fi and one from a wired computer are not a clean before-and-after comparison.
Different services can use different routes, servers, stream counts, durations and accounting. A higher result from another service is not by itself proof that either test is fake. A short HTTP test also includes browser and hosting limitations. Treat it as a sample of one path at one time, not a guarantee for every application.
Your three-run worksheet
Enter three completed download results in Mbps. Leave out failed measurements, but note those failures separately. No example values are prefilled. Nothing entered here is uploaded or saved automatically.
Enter your measurements to calculate a summary.
Read the median together with the range
The median is the middle of three sorted values. It reduces the influence of one unusually fast or slow run, but three runs are still a small sample. Keep the range and failures visible rather than presenting the median alone as a stable service guarantee.
Illustrative example: 42, 48 and 90 Mbps have a median of 48 Mbps and a range of 42–90 Mbps. The 90 Mbps run is real in this fictional example, but selecting only that run hides the variation. Repeat at another time under matching conditions before drawing a conclusion.
Do not average percentages from samples with different numbers of attempts. For request loss, combine failed requests and total attempts first, then divide. If you only have rounded percentages and no sample sizes, keep the runs separate.
Estimate a transfer time
Enter the file size in decimal GB and the relevant measured throughput: download for receiving a file, upload for sending one. This is a best-case constant-rate calculation, not a new measurement. The destination server, Wi-Fi variation and protocol overhead may make the actual transfer longer.
Enter a file size and measured speed.
Formula: seconds = GB × 8,000 ÷ Mbps. A decimal gigabyte is one billion bytes; each byte contains eight bits. If your file manager reports GiB, convert it first: 1 GiB is approximately 1.074 GB. Mbps and MB/s are not interchangeable.
Illustrative example: 1 GB at a constant 20 Mbps gives 400 seconds, or 6 minutes 40 seconds, before any additional overhead or slowdown.
Keep a record you can use
Notes remain only in this open page. Print or save a PDF before leaving if you want to keep them. Review what you include before sharing it; omit IP addresses, account numbers and passwords.
One report should answer: what was wrong, when it happened, how you tested, and what changed. It should not claim a router or provider caused the problem unless the evidence establishes that.