LineaChiara

Internet slow at night: compare peak hours without guessing

A connection that slows down every evening deserves a comparison under similar conditions. The time of day alone does not prove provider congestion: backups, streaming, Wi-Fi and the test destination may change too.

By PHABDEV ·

Prepare a fair comparison

Use the same device, connection and test server. Connect directly to the router over Ethernet if possible. Keep the test duration, parallel streams and mode unchanged. Record what other devices are doing: a streaming television or backup may matter even when your test computer is idle.

Compare a period when everything works with the troublesome period. Repeat on different days if possible. Do not change DNS, router, test server and Wi-Fi settings together: you would not know which change made a difference.

Which results to keep

  • Download and upload with the date, time and time zone.
  • Idle latency and latency during transfers, to show how responsiveness changes under load.
  • The test server, connection type and declared competing traffic.
  • Incomplete tests and failed requests, without turning unavailable measurements into zero Mbps.
  • The actual problem, such as “the video stops every minute”, as well as the numbers.

LineaChiara’s local history lets you select and compare two sessions. For a long observation, use Stability only and run targeted speed tests at useful times. Repeated heavy transfers consume data and can worsen the connection you are observing.

What the pattern suggests

If only Wi-Fi gets worse, the comparison does not yet support a fault on the external line. If only one destination slows down, consider that server or the route to it. If multiple services and destinations slow down over Ethernet at recurring times without known household traffic, external congestion is a hypothesis worth asking the provider to investigate.

A large latency rise during upload can be consistent with data waiting in queues. It does not identify where those queues are. Low throughput with stable latency may describe a different bottleneck: avoid diagnosing from one indicator alone.

Make a specific support request

State which time windows you compared, how many days your observations cover and which conditions you kept constant. Ask support to check the line and possible congestion at those times, with your reports attached. A repeatable comparison is useful evidence, but is not itself certification of a contract breach.

Before upgrading your plan or buying hardware, check that the bottleneck is not your local connection or device. Advertised service speeds and an HTTP browser measurement describe different things.

Put the checks into practice

Open the free test and choose your settings. Diagnostic traffic starts only when you press Start. No registration or required fields.

Open free diagnostics

Method and references

LineaChiara measurement method and limitations

These checks help narrow down a problem; they are not a guarantee of its cause. Suggest a correction.

Related guides