What Is a Latency Test? Ping and Lag Explained

A latency test measures how long a small piece of data takes to travel from your device to a server and back again. The answer is reported in milliseconds (ms), and unlike download or upload speed, a lower number is better. It is the same measurement people call ping, and it is the number that decides whether a video call feels natural, whether a game feels responsive, and whether a web page appears the instant you click.

Most speed tests, including the speed test on this site, run a latency test automatically before measuring bandwidth. That is why your result has a "ping" or "latency" figure next to the Mbps numbers.

What a latency test actually does

A latency test sends a tiny request to a server — small enough that the size of your connection makes almost no difference — and starts a stopwatch. When the reply arrives, it stops the stopwatch. That round trip is your latency, usually written as RTT (round-trip time).

The test normally repeats this several times and reports the best or the median result, because any single measurement can be delayed by a moment of Wi-Fi interference or a busy CPU. A single sample tells you very little; a handful of samples tells you the shape of your connection.

What is a good latency? Numbers in milliseconds

Those numbers assume an idle connection. Latency measured while the line is busy is a different — and often much more revealing — figure. More on that below.

Latency, ping, jitter and lag: the same family, different words

• Latency — the general term for delay, measured in milliseconds.
• Ping — both the name of the classic command-line tool and the everyday word for the latency figure it produces. In practice, "ping" and "latency" mean the same thing in a speed test result.
• Jitter — how much the latency varies between one measurement and the next. Steady 60 ms is far better for a call than latency bouncing between 20 ms and 200 ms. Jitter under about 30 ms is generally fine for voice and video.
• Lag — the subjective experience caused by high latency, high jitter, or packet loss. Lag is what you feel; latency is what you measure.

Download latency and upload latency

Some tests — including the broadband labels that providers publish — report latency in more than one place, which raises a reasonable question: is there a separate download latency and upload latency?

A single round trip involves both directions, so basic latency is one number covering the outbound and return journey together. What newer tests report separately is latency measured while the connection is loaded: ping during a heavy download, and ping during a heavy upload. Those two figures can be very different from each other and from the idle number.

A typical pattern looks like this:

When loaded latency is many times the idle figure, the cause is almost always bufferbloat — oversized queues in your router or modem holding small time-sensitive packets behind bulk traffic. It is a common, fixable fault, and an ordinary latency test taken on an idle line will never show it.

How to run a latency test properly

1. Test on a wired connection first. Wi-Fi typically adds 5–30 ms of its own and makes results noisier. If wired latency is good and wireless is bad, the problem is your Wi-Fi, not your line.
2. Close what else is running. A cloud backup or a game update in the background will inflate the result and hide the connection's true idle latency.
3. Run it more than once. Take three or four measurements over a few minutes and note the range, not just one number.
4. Then deliberately load the line. Start a large download or upload and run the latency test again. The gap between idle and loaded latency is the single most useful diagnostic most people never look at.
5. Test at different times of day. Shared-medium connections such as cable often show higher latency and jitter during the evening peak.

If your figures move around a lot between runs, that is normal to a point — speed test results vary for reasons that have nothing to do with a faulty connection.

What makes latency high

• Distance. Light in fibre covers roughly 200 km per millisecond, and the route is never a straight line. A server on another continent will always cost you 100 ms or more, and no upgrade changes physics.
• Connection technology. Fibre is the lowest; cable adds a few milliseconds; DSL more; geostationary satellite adds 500 ms or more by design. See fiber vs cable vs DSL for what each technology can realistically deliver.
• Wi-Fi. Interference, distance from the router and older standards all add delay and jitter.
• Congestion and queuing. Either at your own router (bufferbloat) or in the provider's network at peak times.
• The device itself. An overloaded phone or laptop can add tens of milliseconds before a packet ever reaches the network.

How to lower your latency

• Use Ethernet for anything latency-sensitive — gaming consoles, work laptops, desk phones.
• Enable Smart Queue Management (SQM/fq_codel) on your router if it offers it. This is the direct fix for latency that spikes under load.
• Move or upgrade the router if wireless latency is the problem — the same steps that improve internet speed generally cut latency too.
• Pick nearer servers where the application lets you, such as a game region or a video-conference data centre.
• Do not buy a faster plan to fix latency. More bandwidth does not reduce round-trip time. If the line is already fast enough for your household, upgrading changes almost nothing about how responsive it feels.

In short

Run the speed test to see your own latency alongside download and upload, then read what those three numbers mean together.

Read next

• Download, upload and latency explained — how the three metrics relate to each other.
• Bufferbloat explained — why latency spikes the moment your connection gets busy.
• What is a good internet speed? — the bandwidth side of the same question.