A speed number is only a starting point. The same symptom can come from the device, Wi-Fi, local traffic, broadband line, provider route, DNS or the service itself. Use controlled comparisons to find the boundary where performance changes.
Test under known conditions
Pause large transfers and run three tests on the same device. Note the time, room, connection type and whether a VPN is active. The browser test uses real HTTPS transfers and measures the complete path to LumaProbe, not only the provider line.
Change one part of the path
Compare beside the router with the problem room, then Wi-Fi with Ethernet where possible. If wired performance is healthy but a room is poor, investigate coverage. If every device is poor on both, check official provider information and the access connection.
Interpret more than download
Upload affects calls and backups; latency affects responsiveness; jitter shows variation. Failed requests are unsuccessful HTTPS samples, not a formal ICMP packet-loss percentage. Repetition matters more than one unusually good or bad run.
LumaProbe requires no account, stores public test history only in your browser, publishes its implemented method and links to official provider checkers instead of inventing live outage claims.