Find the lag, then fix it

One test, about 30 seconds: ping, jitter, download, upload, packet loss, and how much your ping jumps when the line is busy. Then a plain verdict for gaming, video calls and streaming, and the one thing to fix first. Free, and nothing is stored.

The test needs JavaScript.

Ping
–
Jitter
–
Download
–
Upload
–
Packet loss
–
Bufferbloat grade
–

What the test measures

  • Ping: how long a small request takes to reach the nearest Cloudflare location and come back, in milliseconds. We send about 20 and show the median. Lower is better; under 50 ms suits competitive games. What's a good ping?
  • Jitter: how much ping changes from one request to the next. Steady matters as much as low: jitter above about 30 ms makes calls stutter. Acceptable jitter
  • Download and upload: how fast data moves each way, in megabits per second. The test grows the transfers until each takes about a second, so it measures fast lines without wasting data on slow ones. It tops out near 1 Gbps.
  • Packet loss: the share of small UDP messages, the kind games and calls use, that never arrive. On a healthy line it's 0%. How much is too much
  • Bufferbloat: how much your ping rises while the line is busy with a download or upload. This is the lag you feel when someone else in the house starts a download. We grade it from A to F. The grades explained

Reading your verdict

The verdict checks your numbers against what each use needs: competitive gaming, video calls and 4K streaming. Speeds come from what the services themselves recommend: Netflix asks for 15 Mbps for 4K, and Microsoft Teams' best-quality video uses about 4 Mbps each way. Latency, jitter and loss limits are our own thresholds, explained on how the test works.

Each use gets a tick or a cross with the reason, such as "jitter 38 ms: calls will stutter". Below that, one link: the fix for your weakest result. Packet loss comes first, then bufferbloat, then jitter, then ping, then speed, because that's roughly the order in which they spoil games and calls.

Make the result count

  • Test the way you play. If you game on Wi-Fi, test on Wi-Fi. If you want to know what your internet plan delivers, test on a cable, close to the router.
  • Test during the problem. Lag at 9 pm and a clean test at noon are both real. Run it when things go wrong.
  • Run it twice. One bad run can be bad luck. Two in a row is a pattern.
  • Pause other traffic if you want your line's best case, or leave it running if you want to see what a busy evening does to your ping.

Why your numbers can differ from other tests

Every speed test measures to a different server with a different method. Ours measures to the nearest Cloudflare location over ordinary web requests, the way your browser and most apps talk to the internet. Speedtest by Ookla picks a nearby server from its own network and opens several connections at once, so it often shows higher speeds. Neither is wrong; they answer different questions. Why results differ

What it can't see

A browser can't send the "ping" that a command line sends (ICMP), and it can't reach game servers directly. So our ping is to Cloudflare, usually a few milliseconds from you, and your ping to a game server will be higher, by however far away that server is. We say so on every result. For a specific game's servers, check the game's own network stats overlay.

Focused tests

Each of these runs a shorter version with its own explanation:

Privacy

The test sends data only to our own server on Cloudflare. We don't store your results, your network name or your IP address. Analytics, once it's on, records that a test ran and coarse bands like "ping 20–50 ms", never your numbers. Details are in our privacy policy.