Throughput is how much data moves per second, latency is how long a round trip takes, and jitter is how much that delay wobbles. A network can be great at one and poor at another, which is why "fast" is never a single number.

When people describe a connection as fast or slow, they usually mean one specific thing without realizing there are three separate qualities in play. A video call can stutter on a gigabit line, and a retro game can feel razor-sharp on a modest one. Understanding throughput, latency, and jitter as distinct measurements — each with its own units, causes, and fixes — is the difference between guessing at network problems and actually diagnosing them.

Throughput: how much data per second

Throughput is the volume of data your connection can actually move in a given amount of time. It is what most speed tests headline, reported in megabits per second (Mbps) or gigabits per second (Gbps). Think of it as the width of a pipe: a wider pipe pushes more water through per second.

Throughput is the metric that matters for anything involving moving a lot of data continuously — downloading a large game, streaming 4K video, copying files to a NAS, or backing up a laptop. If a 4K stream buffers, or a file copy takes forever, throughput is usually the suspect.

A crucial and widely misunderstood point: throughput is measured in bits, but your file manager reports copies in bytes. Since there are 8 bits in a byte, a 100 Mbps link tops out around 12.5 MB/s, not 100. If that ratio surprises you, our guide to Mbps vs. MB/s and bits vs. bytes spells out the math with worked examples. Confusing the two makes a perfectly healthy connection look eight times slower than it is.

Latency: how long the round trip takes

Latency, often called ping, is the time it takes for a small piece of data to travel to a destination and back, measured in milliseconds (ms). If throughput is the pipe's width, latency is how long it takes a single drop to make the trip.

Latency has almost nothing to do with throughput. You can have enormous throughput and still suffer high latency, and vice versa. A satellite internet link, for instance, might deliver plenty of throughput while imposing 600+ ms of latency simply because the signal travels to orbit and back.

Latency is what governs how responsive a connection feels. It determines the lag between clicking and seeing a result, the delay before a game registers your input, and the awkward pauses in a phone or video call. For these interactive uses, low latency matters far more than raw throughput. A competitive gamer would happily trade half their throughput for 20 ms lower ping.

Several things add to latency: physical distance to the server, the number of network hops along the way, processing delays in routers, and — critically — congestion on your own network. That last cause is worth its own section.

Jitter: how much the delay wobbles

Jitter is the variation in latency over time, also measured in milliseconds. If your ping bounces between 15 ms, 90 ms, 20 ms, and 110 ms, you have high jitter even though the average might look acceptable. Consistent latency, even if slightly higher, is often better than latency that swings wildly.

Jitter is the quiet culprit behind choppy video calls and garbled VoIP audio. Real-time media has to reassemble a steady stream of packets in order and on time. When packets arrive at unpredictable intervals, the receiver either waits (adding delay) or drops them (causing glitches and dropouts). This is why a call can sound terrible even when a speed test reports healthy numbers — the test measured average throughput and latency, not the moment-to-moment consistency.

How they interact

The three metrics are independent enough that you must consider all of them, but they do influence each other in one important way: saturating your throughput often wrecks your latency and jitter. When you max out your upload — say, a large cloud backup — packets pile up in a buffer waiting their turn, and every other connection's latency spikes. This effect is called bufferbloat, and it is why a video call falls apart the moment someone in the house starts a big upload. The fix is not more throughput; it is smarter queue management on your router.

This interaction explains a lot of everyday frustration. Your connection is not "randomly" bad during calls — it is predictably bad whenever something else is consuming the pipe.

Which metric matters for which task

Metric	Unit	What it affects most	Typical "good" value (home, as of 2026)
Throughput	Mbps / Gbps	File copies, downloads, 4K streaming, backups	Enough to cover your needs; 100+ Mbps comfortable, gigabit generous
Latency (ping)	milliseconds (ms)	Gaming, video calls, responsiveness, web page snappiness	Under ~20 ms to nearby servers is excellent; under 50 ms is fine; 100+ ms is felt
Jitter	milliseconds (ms)	VoIP, video calls, live streaming quality	Under ~5 ms is excellent; under 30 ms usually acceptable; over 50 ms is trouble

The takeaway from this table: a household streaming movies cares mostly about throughput, while a household full of gamers and remote workers on video calls cares mostly about latency and jitter. The "right" connection depends on what you actually do.

How to measure each

• Throughput. Any speed test reports it. For internet, use an online tester; for your own network, use a local network speed test so you measure device-to-device speed rather than your ISP link. The two are different measurements, as our LAN vs. internet test guide explains.
• Latency. The classic tool is ping, available on Windows, macOS, and Linux. Run ping 8.8.8.8 and watch the times; the round-trip figure in milliseconds is your latency to that host.
• Jitter. Run a continuous ping (for example, ping -n 50 on Windows or ping -c 50 on macOS/Linux) and observe how much the times vary. Many speed-test tools also report a jitter figure directly. The wider the spread, the worse your jitter.

Putting it together

Next time something on your network feels off, resist the urge to reach for a single speed number. Ask which of the three qualities the symptom points to. A slow download is a throughput question. A laggy game is a latency question. A choppy call is almost always a jitter or bufferbloat question. Naming the right metric is most of the diagnosis, and it keeps you from throwing money at a faster plan that fixes nothing.

Throughput, latency, and jitter together describe a connection far more honestly than any one of them alone. Learn to read all three, and "the network is slow" becomes a specific problem you can actually solve.