Network Diagnostic Methodology

We believe a speed test is only useful if you understand exactly what it measures. This page documents how our diagnostic engine calculates each metric so you can interpret your results with confidence.

Where the test runs

The entire measurement runs client-side in your browser using standard Fetch and XMLHttpRequest networking. Our servers act only as lightweight byte sources and sinks — there is no heavy backend compute — which keeps the test fast and privacy-respecting. Measurement endpoints are served from South African hosting nodes to reflect realistic in-country latency rather than distant international round trips.

Latency (ping) and jitter

We issue a sequence of small, uncached requests to a dedicated latency endpoint and record the round-trip time of each using the browser's high-resolution timer. The first sample is discarded to remove connection warm-up bias. Ping is reported as the best (lowest) stable round trip, and jitter is the mean absolute variation between consecutive samples — a measure of latency stability that matters for gaming and video calls.

Download throughput

We open several parallel connections to a streaming endpoint that emits incompressible random data, then read the byte stream and total the bytes received over a fixed measurement window. Speed is calculated as:

• Mbps = (bytes × 8) ÷ seconds ÷ 1,000,000

Parallel connections are used because a single TCP stream rarely saturates a modern fibre or 5G line; multiple streams reflect real-world usage such as streaming, downloads and browsing happening at once.

Upload throughput

Upload is measured by posting incompressible payloads to a sink endpoint over parallel connections, using upload progress events to track bytes transferred over the measurement window. The same Mbps formula applies.

Why your result may differ from your line speed

• Wi-Fi interference, distance from the router, and older hardware.
• Background apps and other devices consuming bandwidth.
• Peak-time congestion on your ISP or the wider network.
• Browser, CPU and device limitations on very high-speed lines.

For the most accurate reading, test over a wired connection with no other downloads running. Results are indicative and intended for comparison over time, not as a contractual measurement.

Ready to benchmark your line? Run the speed test →