First, compare like with like
A speed test showing 100 Mbps and a download showing 12 MB/s are not necessarily disagreeing. Lowercase b means bits; uppercase B means bytes. Eight bits make one byte, so 100 ÷ 8 = 12.5 MB/s is the theoretical equivalent.
Write down the complete unit beside each number. If your downloader offers a bits-per-second display, use it for the comparison. A smaller number alone is not evidence that your provider is withholding speed.
Work out a useful waiting time
For a decimal 10 GB file, there are 10,000 MB to transfer. At a constant 12.5 MB/s, the calculation is 10,000 ÷ 12.5 = 800 seconds: 13 minutes 20 seconds. This assumes the whole advertised rate carries file data without interruption.
For planning, use a sustained rate from the actual download. At 8 MB/s, the same file needs 1,250 seconds, or 20 minutes 50 seconds. Once downloading has begun, divide the remaining size, not the original size, by the observed rate. Both results are estimates, not promises.
Do not silently swap GB and GiB
GB and GiB represent different quantities: 1 GB = 1,000,000,000 bytes; 1 GiB = 1,073,741,824 bytes. A 10 GiB transfer therefore contains about 7.37% more data than a 10 GB transfer. At the same rate it also takes about 7.37% longer.
Use the unit stated by the application, or its exact byte count if the label is unclear. Changing the unit changes the number, not the file. Network traffic, downloaded archive size and final installed size are separate measurements; do not substitute one for another.
Find the bottleneck before buying more speed
Pause competing transfers and compare the same download over Wi-Fi and a wired connection where practical. Check the application's own speed limit. Record the unit, sustained rate and connection type so each comparison answers a specific question rather than introducing several changes at once.
A busy source server, shared connection or local device can limit delivery. Protocol traffic also uses capacity, so there is no universal percentage to subtract. If a speed test is fast but one download remains slow, investigate that transfer before upgrading your plan. Allow separate time for verification, extraction or installation after the transfer finishes.