How to Test Internet Speed Accurately

Why Accurate Speed Testing Matters

A speed test taken carelessly can give wildly misleading results — either flattering results that hide a problem, or artificially low results that make you call your ISP unnecessarily. Accurate testing requires controlling variables that most users don't consider.

Speed tests are valuable for several real-world purposes:

• Verifying you're receiving speeds close to your ISP's advertised rates
• Diagnosing whether slow performance is a bandwidth problem or a latency/routing issue
• Identifying time-of-day congestion patterns
• Comparing WiFi performance to wired performance
• Troubleshooting upload bottlenecks for video calls and cloud sync
• Documenting speeds before and after changes (new router, ISP plan change)

Start with our speed test for a quick measurement, then use the methodology below for accurate, repeatable results when you need reliable data.

Preparing for an Accurate Test

Before running any speed test, eliminate background noise that will artificially reduce your measured speeds:

• Close all applications that access the internet: streaming services, cloud sync (Dropbox, Google Drive, OneDrive), software updaters, antivirus updates, and browser tabs with auto-refreshing content
• Disconnect or pause other devices on the network, or at minimum ensure no other device is actively downloading or streaming
• Use a wired Ethernet connection for the most accurate results. WiFi introduces variable overhead that makes results inconsistent between runs
• Test multiple times and at different times of day — run tests at 10 AM, 6 PM, and 10 PM to capture both off-peak and peak-hour performance
• Test against multiple servers including your ISP's own speed test server (if available), a nearby neutral server, and a distant one

A single speed test run is a snapshot, not a measurement. Three or more runs at different times reveal your actual connection quality.

Choosing the Right Speed Test Tool

Different speed test tools use different methodologies and infrastructure, which is why the same connection can test at 400 Mbps on one tool and 180 Mbps on another:

• Our speed test — browser-based, no installation required, measures download, upload, and latency
• Speedtest.net (Ookla) — the most widely used; uses a single-stream TCP connection which may underestimate speeds on high-latency connections; server closest to you is auto-selected
• Fast.com (Netflix) — specifically tests against Netflix's CDN infrastructure; useful for diagnosing Netflix buffering issues specifically
• Waveform Bufferbloat Test — uniquely tests latency under load, diagnosing bufferbloat that standard tests miss entirely
• iPerf3 — open-source command-line tool for professional network testing; requires access to a server running iPerf3 on the other end
• nperf.com — runs multiple parallel streams, which gives more accurate results on high-latency intercontinental connections

For most users, running our speed test alongside Speedtest.net covers the bases. If results differ significantly between tools, the difference often reveals whether your ISP is optimizing traffic to specific speed test servers (a practice called "speed test optimization").

Interpreting Your Speed Test Results

A speed test typically reports three core metrics:

• Download speed (Mbps) — how fast data flows from the internet to your device. Most relevant for streaming, downloads, and web browsing.
• Upload speed (Mbps) — how fast data flows from your device to the internet. Critical for video calls, cloud backup, and livestreaming.
• Ping / latency (ms) — the round-trip time to the speed test server. Relevant for responsiveness in gaming and video calls. Note that this is ping to the speed test server, not necessarily the server you care about — use our ping test to test specific hosts.

Some advanced tests also report:

• Jitter (ms) — variation in latency; under 10 ms is excellent
• Loaded latency — ping measured while the speed test is running (the Waveform test); reveals bufferbloat if much higher than idle latency

Compare results to your plan: a 500 Mbps plan should deliver 400–500 Mbps on a wired test at off-peak hours. If you're consistently getting less, contact your ISP with documented results.

Testing WiFi vs Ethernet and Diagnosing Differences

One of the most valuable comparisons is testing your speed on WiFi versus a wired Ethernet connection. This quickly identifies whether your WiFi is a bottleneck:

# Test sequence:
1. Run speed test on Ethernet (direct connection to router)
2. Note results
3. Disconnect Ethernet, test on WiFi in same location
4. Note results and compare

Common findings and what they indicate:

• Ethernet matches plan speed, WiFi is much lower — WiFi is the bottleneck; consider upgrading router, changing channel, or adding a WiFi 6 access point
• Both Ethernet and WiFi are below plan speed — router or ISP issue; try connecting directly to modem
• Ethernet at full speed, WiFi at full speed — no bottleneck; seek other causes for perceived slowness (latency, specific application issues)
• Results vary dramatically between WiFi test runs — WiFi interference or signal instability; check 2.4 vs 5 GHz bands separately

For thorough diagnosis, combine speed testing with a latency test and traceroute to build a complete picture of your connection's health.

Frequently Asked Questions