Methodology

How Laju measures your connection

Four steps, published parameters, and an honest list of what the result cannot tell you.

  1. 01

    1. Latency and jitter

    The browser sends a series of tiny, uncacheable requests. The fastest round trip becomes the latency figure, while jitter is the mean difference between consecutive probes. This is HTTP rather than ICMP, so the number reads slightly higher than a ping command.

  2. 02

    2. Download

    Several parallel connections pull random data from our server and bytes are counted as they land. Random data is used because it cannot be compressed in transit, so nothing inflates the figure.

  3. 03

    3. Latency under load

    While the download runs, latency is measured again. A large increase points to over-full buffers somewhere along the path — the usual reason for lag on a line that looks fast.

  4. 04

    4. Upload

    Random data is sent back to the server and discarded on arrival. Browsers report upload progress slightly ahead of the wire, which is why the upload warm-up window is the longer of the two.

Parameters

The numbers the engine runs on

These values are read straight from the measurement configuration, so this page always matches what actually runs.

Latency probes
14
Download phase duration
11 seconds
Upload phase duration
11 seconds
Parallel download connections
5
Parallel upload connections
4
Warm-up discarded (down / up)
1.6 / 2.2 seconds

Limits

What this result cannot tell you

We would rather state the limits than let a number claim more than it can support.

  • The test runs against a single server, not a global network. The path to other services may look different.
  • Measurement happens inside a browser over HTTPS, so there is encryption and processing cost a system-level tool would not pay.
  • Wi-Fi, tired cabling or an ageing router is often the slowest link, rather than the subscription itself.
  • A VPN, proxy or network-level ad blocker changes the result because traffic takes an extra path.
  • Other devices using the same connection during the run will pull the numbers down.
  • Upload progress is reported by the browser from its socket buffers, so the first seconds tend to look faster than reality.

Common questions

Things people ask

Why does this differ from other speed tests?

Every tool uses different servers, durations, connection counts and averaging rules. Differences of tens of percent between tools are normal. A very large gap usually means the path itself is different — the other tool may test against your operator's nearest server while Laju tests against its own. And when Laju is served from a local server, the payload never leaves your machine at all, so the number is not an internet speed. Comparing results from the same tool over time is far more useful.

Why is my IPv6 address not detected?

Usually because your network or provider has not enabled IPv6, or a device along the path blocks it. It is not a fault on your machine; it simply means your traffic still travels over IPv4.

What does jitter mean?

Jitter is how much the round-trip time changes from one probe to the next. Low jitter means a steady stream; high jitter makes voice and video sound choppy even when throughput is large.

What is latency under load?

It is the latency measured while the line is saturated by a download. If it rises far above the idle figure, there are probably over-full buffers in your router or in the provider network.

How much data does a test use?

It depends on your speed, because the test runs for a fixed time rather than a fixed size. A faster connection moves more data in the same window. The exact amount is shown in the details of every result.

Are my results stored on your server?

No. The test payload is random bytes that are discarded immediately, and results are kept only in your browser's local storage. You can erase them at any time from the History page.

How do I get the most representative result?

Close active downloads and streams, use a cable if you can, move closer to the router on Wi-Fi, then run the test a few times at different hours and compare the pattern.

Ready to see how your connection is doing?

The test takes about half a minute and asks for no personal details at all.