Measure download, upload, ping and jitter — then find out what the numbers actually mean
Takes about 20 seconds. Runs entirely in your browser — no app, no plugin, nothing to install.
Measure download, upload, ping and jitter — then find out what the numbers actually mean
Takes about 20 seconds. Runs entirely in your browser — no app, no plugin, nothing to install.
In March 2024 the FCC raised its benchmark for fixed broadband to 100 Mbps download and 20 Mbps upload, up from the 25/3 standard that had stood since 2015. That is the closest thing to an official answer. The more useful answer depends on how many people share the line and what they do at the same time.
| Download speed | What it comfortably supports |
|---|---|
| Under 5 Mbps | Browsing and email. Video buffers, calls struggle. |
| 5 to 25 Mbps | One person. HD streaming, one 4K stream at the top of the range. |
| 25 to 100 Mbps | A couple or small family. Several HD streams, calls, normal downloads. |
| 100 to 500 Mbps | A full household. Multiple 4K streams, gaming and large downloads together. |
| Above 500 Mbps | The connection stops being the bottleneck. The limit moves to your Wi-Fi, your device or the remote server. |
Mbps is megabits per second. File sizes are in megabytes, and there are eight bits in a byte. That single conversion causes more confusion about internet speed than anything else: a 100 Mbps connection downloads at about 12.5 MB/s at its theoretical best, not 100 MB/s. When your browser shows a file downloading at 11 MB/s on a 100 Mbps line, nothing is wrong — that is the line running at full speed.
The quick conversion: divide Mbps by 8 to get megabytes per second. 50 Mbps is about 6 MB/s. 400 Mbps is about 50 MB/s. A gigabit connection tops out near 125 MB/s.
Add up what runs at the same time, using the figures the services themselves publish. Everything else is guesswork.
| Activity | Recommended |
|---|---|
| Netflix in 4K | 15 Mbps per stream |
| Netflix in 1080p | 5 Mbps per stream |
| YouTube in 4K | About 20 Mbps |
| Zoom, 1080p call | About 3.8 Mbps each way |
| Online gaming | Under 5 Mbps, but low and steady ping |
| Cloud backup | 10 Mbps upload or more |
Two 4K streams and a video call comes to roughly 35 Mbps. Gaming is the surprise on this list: it needs almost no bandwidth, but it is the activity most sensitive to ping and jitter.
Ping is how long a packet takes to reach a server and come back. It is independent of speed — a gigabit satellite link can have terrible ping, because much of it is simply the distance travelled. Under 20 ms feels instant, under 50 ms is comfortable for gaming, and past 150 ms the delay becomes noticeable in anything interactive. Every browser test measures to its own server, so part of the figure is simply the distance to it — which is why a result is most useful compared against itself over time.
Jitter is how much that ping varies from one moment to the next, and it is the number that explains the most common complaint of all: my speed test looks fine but my calls keep breaking up. Voice and video need packets to arrive evenly spaced. When they arrive in bursts, the call clips words and freezes even though the average speed is perfectly good. Under 5 ms is very stable; above 30 ms you will hear it.
A speed test measures the entire path from your device to the server, and Wi-Fi is usually the weakest link in it — not the line you pay for. The common causes, in roughly the order they bite:
The diagnostic that settles it takes two minutes: run the test over Wi-Fi, then run it again with an Ethernet cable. If the cable is much faster, the problem is inside your home and no provider can fix it for you.
The test opens several connections at once to the nearest edge server and moves real, random, incompressible data over them — random because a block of zeros would compress in transit and report a speed your line cannot actually deliver. Bytes are counted as they arrive rather than assumed from what was requested, so a truncated or retried request cannot inflate the result.
The first stretch of every measurement is discarded. A TCP connection does not start at full speed, it ramps up, so timing from the very first byte understates a fast connection badly — and the faster the line, the worse the understatement. Latency is reported as the lowest of fourteen round trips, because a single sample can only be inflated by a busy moment, never reduced below the true path time. Nothing is installed, and no file from your device is ever sent.