Latency to our servers

This measures HTTP response time, not an ICMP ping: ICMP is not available from a browser. Our number is always slightly higher than what the ping command shows. The first request is discarded — it includes setting up the connection rather than the latency of the channel.

Only one measurement point is available so far, so there is nothing to compare against. More points will appear as they are connected.

How latency differs from speed

Speed is how much data passes per second. Latency is how long data takes to travel one way and back. They are different quantities, and high speed does not rescue you from high latency: a wider road does not make it shorter.

Downloading a file needs speed; a conversation or a game needs low latency. A hundred-megabit channel with three hundred milliseconds of latency downloads beautifully and calls badly: people start talking over each other because they hear the reply almost a third of a second late.

Why we show the minimum, not the average

Latency can be increased by a queue in the network, a busy Wi-Fi, power saving on a phone — by anything at all. Nothing can decrease it: a signal will not outrun the speed of light in the cable.

So the best of the samples is the most honest estimate of the real travel time, while the average is always inflated by random interference. The median beside it shows how typical that best result is: if it sits close to the minimum, the channel is stable.

What spread is and why it matters more

Spread is how much the samples differ from one another. A steady ninety milliseconds is noticeably nicer than jumps between forty and two hundred at the same average.

The reason lies in how voice and video work: they gather the stream into a buffer sized for the worst case. A steady channel lets that buffer stay small; a jumpy one forces it to grow, and the conversation starts lagging. When the spread is comparable to the latency itself, the cause is usually Wi-Fi or a congested network rather than the server.

What affects latency and what to do about it

Distance contributes most: light travels about two hundred kilometres per millisecond in fibre, and a round trip to another continent physically cannot be quick. Then come the number of intermediate hops, channel load, and the quality of the last mile — most often the home Wi-Fi.

Hence the order of actions. Rule out your own side first: a cable instead of Wi-Fi, closer to the router, fewer active downloads. Then compare the measurement points with each other — if a near one answers quickly and a distant one slowly, that is ordinary physics, not a fault.

Frequently asked questions