Why Is My Starlink Slow or Unstable Every Evening?

Why does my Starlink speed drop every evening?

Starlink speed often drops in the evening because more people are using the network at the same time. Starlink defines peak usage as 6 PM to 9 PM local time, when network load is greatest, while the Australian Competition and Consumer Commission measured busy weeknight hours from 7 PM to 11 PM.

The data resolves an apparent contradiction: a connection can be broadly fast and still feel worse after dinner. In its December 2024 satellite-broadband measurements, the Australian Competition and Consumer Commission recorded Starlink average download speed of 192 Mbps across all hours and 165.5 Mbps during busy weeknight hours. Upload performance was much steadier, declining from 28.6 Mbps to 27.8 Mbps. That pattern is consistent with evening demand affecting download capacity more visibly than upload capacity.

Do not use an isolated high result as the standard against which every evening is judged. The ACCC observed peak Starlink results of roughly 470 Mbps download and 74 Mbps upload; those are highs, not guaranteed floors. Starlink’s own service specifications list fixed Standard service at 25–100 Mbps download, 5–10 Mbps upload, and 25–60 ms latency, while stating that location, time of day, and plan precedence affect actual performance.

The practical diagnosis is blunt: if service is stable during the day and slows in a repeatable evening window, congestion is a credible explanation. If the connection drops, freezes, or shows interruptions regardless of hour, do not blame congestion first.

How can I tell whether Starlink congestion or an obstruction is causing slow speeds?

You can distinguish congestion from an obstruction by testing the network path in order: dish-to-internet first, then router-to-device, while checking the dish’s sky view. Congestion generally produces a time-linked slowdown; an obstruction produces service interruptions when satellites pass through the blocked part of the sky.

Starlink’s installation guidance is explicit: a dish needs a clear view of the sky as satellites move overhead. A tree branch, pole, or roof can obstruct that connection and cause interruptions. Use the app’s “Check for Obstructions” tool at the installed location or a proposed replacement location. If ground level cannot provide a clear view, Starlink suggests an elevated roof, pole, or wall installation.

This sequence avoids a common error: treating every poor device speed test as proof that the satellite link is slow. Starlink lists the installation environment, cable condition, router and device location, interference, and too many concurrent connections among factors that affect performance. The weak link may be indoors.

Router placement is therefore a diagnostic intervention, not a ritual. Put the router where the devices that need it can connect without unnecessary distance or interference, then compare results before and after. If a larger property needs coverage beyond that location, a mesh or Ethernet-based extension should be judged by the same comparison: does the device result move closer to the dish-to-internet result? For a small workplace, this can be incorporated into routine premises planning alongside work such as professional office cleaning, rather than treated as an unexplained connectivity emergency.

Why is my Starlink ping high at night when gaming?

Night-time gaming ping rises when peak demand affects network pathing and scheduling, but local downloads and Wi-Fi conditions can also contribute. Latency is the round-trip time for data to travel to a destination and back; it matters more to real-time games than a large download speed figure does.

Starlink identifies satellite-to-ground pathing, gateway routing, and fronthaul scheduling affected by the number of users on a satellite beam as latency sources. Its latency report focuses on the 6 PM to 9 PM local peak window for precisely that reason. In the U.S. measurement period described in the Starlink latency report, median peak-hour latency fell from 48.5 ms to 33 ms, while p99 latency fell from more than 150 ms to less than 65 ms. The median describes the middle result; p99 describes the much worse tail of results that can turn a game into a sequence of delays.

The ACCC measured average Starlink latency of 29.8 ms across all hours and notes that lower latency is relevant to real-time applications including gaming. By contrast, a 2024 IEEE International Conference on Communications study, based on nearly 1.7 million crowdsourced measurements from 309 users in 29 countries, reported median round-trip times of 40–50 ms and median peak-time goodput down to 50 Mbit/s. These are not conflicting promises. They are measurements from different datasets, locations, and methods, which is exactly why a household should diagnose its own repeated pattern rather than chase one universal number.

Run the app comparison while gaming is otherwise idle, then repeat it when another device is downloading. Starlink says its router uses active queue management, called fq_codel, intended to prevent one Wi-Fi user’s large download from affecting another user’s gaming latency. If latency still rises only during the same evening period, preserve the dates, times, plan details, app results, and obstruction findings for a support request or a plan review. If the evidence instead shows a blocked view or a local Wi-Fi gap, relocation or home-network work is the appropriate first fix.