A 100 Mbps connection does not mean a file downloads at 100 MB/s. The lowercase b means bits; uppercase B means bytes. Convert the units before deciding whether the download is unusually slow.
- Read the exact units in the plan, test and download app.
- Convert bits and bytes before comparing.
- Run an Internet test with the same device and connection and record the conditions.
- Compare a repeat download from a legitimate source and change one condition at a time.
What the units tell you
NIST defines one byte as eight bits. For decimal units, divide Mbps by eight to obtain MB/s: 100 Mbps becomes 12.5 MB/s. That is a theoretical conversion, not a promise that a real file reaches it.
For a decimal 1 GB file, theoretical seconds are file bytes × 8 ÷ bits per second. At 100 Mbps: 1,000,000,000 × 8 ÷ 100,000,000 = 80 seconds. A GiB is a different size; keep the file’s unit visible rather than silently treating GB and GiB as identical.
A worksheet to use
Use these decimal examples as a calculation check, then record your actual connection and file. Do not compare a single peak reading with an entire download average.
| Item | What to record |
|---|---|
| 100 Mbps | 12.5 MB/s decimal 1 GB in 80 seconds theoretically. |
| 500 Mbps | 62.5 MB/s decimal 1 GB in 16 seconds theoretically. |
| 1 Gbps | 125 MB/s decimal 1 GB in 8 seconds theoretically. |
| Your observation | File size and unit, elapsed time, app unit, source, Wi-Fi or cable, and background transfers. |
Check in this order
- Read the exact units in the plan, test and download app.
- Convert bits and bytes before comparing.
- Run an Internet test with the same device and connection and record the conditions.
- Compare a repeat download from a legitimate source and change one condition at a time.
If the result still looks wrong
Cloudflare explains that a speed test includes the local network, device, path and test server, and is not definitive proof of an ISP’s service. If one source is slow while other downloads work, investigate that source before changing the plan.
Record idle and loaded latency too if calls or games suffer. A high throughput number alone does not explain freezing calls. Avoid repeated large tests on a metered plan without checking their data use.
A worked example
A download showing 10 MB/s is approximately 80 Mbps in decimal units, not 10 Mbps. Compare that with a 100 Mbps plan before diagnosing a tenfold shortfall.
Common questions
Does dividing by eight predict the real completion time?
It gives a theoretical conversion. Server limits, Wi-Fi, other traffic and changing transfer rates can make the actual download longer.
Official sources checked
Official documentation checked on 9 October 2026. Examples are illustrative; menu names can differ by language and software version. No benchmark or firsthand measurement is claimed.