Throughput · latency · jitter · packet loss
Speed Test Metrics Glossary
The words on a speed test result, translated. What each metric measures, what units it arrives in, and which ones matter when you are deciding whether a connection is actually behaving.
Why the vocabulary matters
A number without a definition is just a number
Most speed test results arrive as two big figures and a scattering of smaller ones. Download speed and upload speed get the attention; ping, jitter and packet loss decide whether a video call holds together, whether a game stays responsive, and whether a file transfer stalls halfway. When a result disagrees with what you expected, the smaller numbers usually explain why.
The pages below define each term the way a broadband report or an internet service provider's support script uses it. No jargon-only entries: every definition says what is measured, what units you will see, and what a change in that value actually means for a home connection.
Read this page first if you are comparing plans, reading a provider's claims, or trying to explain a problem to support. Then take your own reading and follow the interpretation guide to put the figures in context.
The metric index
Eight terms cover almost every figure you will meet on a speed test result. Select any entry to open its definition.
Measured in Mbps
Download speed
The rate at which data arrives at your device from the network, averaged across the duration of the test. It is the figure that decides how a stream, a page load, or a large download feels in practice.
Reported in megabits per second. A test server close to your region, a wired connection, and an idle household produce the most representative reading.
Measured in Mbps
Upload speed
The rate at which data leaves your device for the network. Most broadband plans are asymmetric: download is provisioned far higher than upload, which is why a video call can struggle even while a stream plays cleanly.
Watch this figure if your household uploads large files, backs up to cloud storage, or runs several video calls at once.
Measured in ms
Ping and latency
Ping is the everyday word for the round trip a small signal makes to a server and back; latency is the same delay described as a property of the connection. The two are used almost interchangeably, and both are reported in milliseconds.
Low is better here. A long-distance test server inflates the figure through no fault of your home network, so compare readings taken against servers in the same region.
Measured in ms
Round-trip time
Round-trip time, usually shortened to RTT, is the strict measurement behind the word ping. It is the time between sending a request and receiving the reply, measured end to end. It includes every hop the packet passes through, not just the last leg inside your home.
Because it counts the whole journey, RTT is the fairest way to compare two connections that exit through different paths or different providers.
Measured in ms
Jitter
Jitter is the amount latency varies from one packet to the next. A connection with a steady 30 ms delay feels smoother than one that swings between 15 ms and 70 ms, even if the average is identical.
It shows up as audio that clips, video that freezes for a beat, or a game that stutters. High jitter points at a congested link, an overloaded Wi-Fi channel, or a route that is losing and retransmitting packets.
Measured as a percentage
Packet loss
The share of data packets that never reach their destination. Transmission protocols resend what goes missing, so a small amount of loss hides behind slightly slower performance rather than a visible error.
Loss is often intermittent. A single clean test does not prove a line is healthy, which is why repeating the test at different times of day tells you more than one reading does.
Measured in Mbps
Throughput
The volume of data actually moved in a given time, as opposed to the maximum the link is capable of. Bandwidth is the width of the pipe; throughput is how much water made it through while you watched.
Throughput is what a file transfer, a stream, or a backup experiences. If throughput sits well below your plan's advertised rate, the gap is where troubleshooting starts.
Measured in Mbps
Bandwidth
The capacity of a connection as provisioned by the provider, expressed as a maximum rate. Advertised plan figures are bandwidth figures; your test results are throughput figures, and the two rarely match exactly.
Every device, cable, and Wi-Fi hop between the wall and the test device shares that capacity, which is why a result measured over Wi-Fi usually lands below the plan number.
Units and scales
Megabits, megabytes, and the factor of eight
Speed test results report megabits per second. Download managers and file sizes report megabytes, and a megabyte holds eight megabits. A 100 Mbps connection moves roughly 12.5 megabytes per second, which is why a plan number and an apparent download rate never look alike.
Latency, jitter and round-trip time are all measured in milliseconds. Packet loss is the odd one out: it is expressed as a percentage of packets that did not arrive. Treating it as a speed figure is one of the most common misreadings of a test result.
One more caution about scale. A 50 Mbps figure measured on a laptop six rooms from the router says more about the Wi-Fi leg than about the line coming into the building. Test what you actually want to know about.
Line states
Four patterns that describe how a connection behaves
Line patterns stand in for connection states throughout this site. They are a reading aid, not a score: the same reading can carry different meaning depending on whether you measured over Ethernet or Wi-Fi, and on what the connection was doing at the time.
Steady reading
A solid rule. One clear measurement on a quiet line, useful as a baseline for later comparison.
Averaged reading
A dashed rule. The test sampled repeatedly and reported a mean, which smooths over short spikes.
Per-packet reading
A dotted rule. Individual packets tracked separately, which is where jitter and loss become visible.
Peak reading
A doubled rule. The highest burst recorded during the test, which is what a large transfer tends to reach.
Reading a result
What to check in the order it matters
A test result has a natural reading order. Start with the fastest-moving figure and work down; the ones at the bottom are the ones people skip, and the ones that usually explain a complaint.
Download and upload speed
Compare against the plan you pay for, then against what you actually do online. Streaming and browsing live on download speed; video calls, cloud backups and large sends live on upload.
Ping and round-trip time
These decide how responsive the connection feels. A high figure measured to a distant server is not necessarily a fault at your end, so repeat the test against a nearby server before drawing conclusions.
Jitter
Look at jitter when the average latency seems fine but calls still break up. A wide spread between the best and worst samples is the tell. Moving a device onto the 5 GHz band, or onto a cable, is the first thing worth trying.
Packet loss and throughput
Any repeatable packet loss deserves a second test before you call anyone; measure throughput during an ordinary evening rather than an idle morning, because the busiest hour is what you actually bought.
Bandwidth against throughput
The gap between the number on your bill and the number on your screen is the useful signal. Note how you measured, where, and on what device, so the comparison means something when you repeat it.
Where the path runs
The number on your screen measures one route, not the whole network. Every hop between your device and the test server adds its own delay, and some of those hops are outside your provider's control.
Terms that get mixed up
Pairs that sound interchangeable in casual conversation but mean different things when you are diagnosing a connection.
Speed or delay
Bandwidth vs. latency
Bandwidth is how much can move at once. Latency is how long one message takes to arrive. A connection can carry a lot of data and still feel sluggish if the delay is long.
Promise or delivery
Bandwidth vs. throughput
Bandwidth is the plan's stated ceiling. Throughput is what your device genuinely moved. Overhead, sharing and weak Wi-Fi all sit in the space between them.
Average or spread
Latency vs. jitter
Latency is the typical delay. Jitter is how much that delay wobbles. Real-time audio and video care more about the wobble than the average.
Questions this glossary gets most often
Is Mbps the same as MB/s?
No. Mbps is megabits per second, the unit speed tests and providers use. MB/s is megabytes per second, the unit file sizes and download managers use. One megabyte is eight megabits, so divide or multiply by eight to compare the two.
My download speed is fine but calls keep breaking up. Which metric should I look at?
Upload speed, jitter and packet loss, in that order. Calls send a steady stream of small packets in both directions, so a narrow upload allowance or an unstable delay shows up as chopped audio long before download speed becomes the problem.
Why does my speed test result change every time I run it?
Because the conditions change. Server choice, time of day, how many devices are active, Wi-Fi signal strength, and what your computer is doing in the background all move the number. Compare readings taken the same way at similar times rather than treating one result as final.
Does a big download speed number mean my connection is good?
It means one part of the connection is performing well. Responsiveness and stability come from latency, jitter and packet loss, and those can be poor on a connection with a very high download figure. Read the whole result, not the headline number.
Should I test over Wi-Fi or with a cable?
Both, and compare them. A wired test tells you what the line can deliver, while a Wi-Fi test tells you what you actually get where you sit. The difference between the two is the cost of the wireless leg and is often the real answer to a speed complaint.
What does a small amount of packet loss actually do?
Missing packets get resent, so the transfer still completes, just slower. The visible effects are hesitation in games, a dip in call quality, or a download that briefly stalls. If loss appears repeatedly across several tests, it is worth raising with your provider.
How this page is built
Definitions written to match real test output
Every entry here describes a figure you can actually see on a speed test result, not a theoretical quantity. Where a term has a stricter technical meaning than the everyday one, both are given so the page stays useful whichever vocabulary the person you are talking to prefers.
The definitions stay deliberately short. If you want the longer walk-through, including what a normal range looks like and how to argue a case with a provider, the interpretation guide takes the next step. For the mechanics of how readings are gathered, see our testing methodology.
Run a test, then read it with the right words
Take a reading on the device and connection you actually use, then bring the terms from this page to it. If a result still does not add up, our team is reachable Monday to Friday, 9:00 AM to 6:00 PM at +17185243042 or TroyRivera@prentissscholey.com, and the office is at 2724 Clarendon Rd, Brooklyn, NY 11226.