A byte contains eight bits, so 100 megabits per second is 12.5 megabytes per second before other transfer overheads.
Why this happens
Broadband packages and many speed tests use megabits per second, written Mbps or Mb/s. A download application may display megabytes per second, written MB/s. The capital B matters. Dividing a megabit figure by eight gives the corresponding byte figure before allowing for protocol overheads and real transfer conditions. A comparison between 100 Mbps and roughly 12.5 MB/s is therefore not automatically evidence that seven-eighths of the connection has gone missing. They can describe the same nominal rate in different units.
Keep the unit beside every number when discussing a playback problem. Also distinguish a test from the router to the internet from a test running on the television over Wi-Fi. Google Nest’s speed-test documentation is explicit about what its router test measures; that result does not automatically describe the final wireless link to a screen in another room. Once the units are clear, compare sustained playback on the actual device. A single headline number still cannot explain a poor source, a decoding problem or an interrupted local connection by itself.
A useful way to work through it
Write down whether each result uses Mbps, Mb/s or MB/s.
Divide megabits by eight only when comparing the same type of throughput measurement.
Identify whether the test ran on the viewing device or only at the router.