- Running tests at several times of day reveals the difference between peak-time contention and an all-day problem.
- Evening peak tests, around 8pm to 10pm, capture the busiest period when contention is highest.
- Wired tests directly to the router give the most accurate picture by removing Wi-Fi as a variable.
- Compare your results against your minimum guaranteed speed, which the provider must meet including at peak times.
A speed test is only as useful as the timing behind it. One result tells you little; a set of results across the day tells you whether your broadband has a real problem and what kind. Building that picture properly is the foundation of any speed complaint, and it takes only a little discipline.
Why timing changes the result
Broadband speed varies through the day, mainly because of contention, the sharing of network capacity that peaks in the evening. A test at a quiet hour may look fine while the same line struggles at nine in the evening. Testing only once, at the wrong time, can either hide a genuine peak-time problem or wrongly suggest an all-day fault. Spreading tests across the day resolves this.
What evening tests reveal
Evening peak tests, roughly 8pm to 10pm, capture the connection under maximum load. If your speed is fine in the morning but drops sharply in the evening, that points to contention. If it is low all day, that points instead to a line fault or a problem independent of busy periods. The contrast between morning and evening is itself diagnostic.
How to test accurately
For results you can rely on, test over a wired Ethernet connection directly to the router, which removes Wi-Fi interference and weak signal as variables. Close other heavy applications during the test, and use a consistent, reputable speed-test tool each time so your results are comparable. Record the date, time and result for each test.
A speed-test schedule for complaints
| Time | What it shows |
|---|---|
| Morning (off-peak) | Best-case speed, less contention |
| Afternoon | Mid-day baseline |
| Evening peak (8-10pm) | Speed under maximum load |
| Repeated over several days | Pattern rather than a one-off |
Comparing to your guarantee
The benchmark is your minimum guaranteed speed, which the provider should meet including at peak times. If your wired evening tests, repeated over several days, fall consistently below that figure, you have evidence of a problem worth reporting. A single low result is easy for a provider to dismiss; a consistent, time-stamped pattern is not.
Frequently asked questions
What is the best time of day to run a broadband speed test?
There is no single best time; run tests at several times, including the evening peak around 8pm to 10pm and an off-peak morning period. The contrast reveals whether a slowdown is peak-time contention or an all-day problem, which is what you need to know.
Should I run multiple speed tests at different times?
Yes. A single test tells you little because speed varies through the day. Testing at several times, repeated over several days, builds the pattern that distinguishes peak-time contention from a genuine fault and provides credible evidence for a complaint.
Why does my speed test result vary so much?
Mainly because of contention, the sharing of network capacity that peaks in the evening, and because Wi-Fi conditions vary. Testing over a wired connection at consistent times reduces the variation and gives a clearer picture of the line's real performance.
Which speed test tool should I use?
Use a consistent, reputable speed-test tool each time so your results are comparable, and test over a wired Ethernet connection to the router. Ofcom's broadband speed resources and established speed-test services are suitable; the key is consistency across tests.
How do I compare my results to my minimum guaranteed speed?
Run wired tests at the evening peak over several days and compare the figures against the minimum guaranteed speed your provider gave you. If the results consistently fall below that guarantee, you have evidence of a problem to report under the Speeds Code.
