How to Test Your Internet Speed Accurately

How to Test Your Internet Speed Accurately

Running a speed test is simple — getting an accurate result takes a bit more care. ISPs sometimes deliver fast speeds specifically to speed test servers, and poor testing conditions can give misleadingly low or high readings. Here's how to test properly.

### Recommended Speed Test Tools

**Speedtest.net (Ookla):** The most widely used speed test. Measures download, upload, and ping (latency). One concern: major ISPs have been accused of optimizing traffic to Speedtest servers, which can inflate results. Use it, but compare with other tools.

**Fast.com (Netflix):** Tests download speed using Netflix's own servers. Since Netflix is a primary traffic driver, this represents real-world streaming performance. Shows upload if you click "Show more info."

**Cloudflare Speed Test (speed.cloudflare.com):** Tests download, upload, latency, and importantly, "loaded latency" — how latency degrades when the connection is busy. This is a more real-world measure of connection quality.

**Google's built-in speed test:** Search "internet speed test" on Google and a quick test appears. Uses Measurement Lab (M-Lab) infrastructure.

**Waveform's Bufferbloat test (waveform.com/tools/bufferbloat):** Tests for bufferbloat — latency increase under load — which causes poor performance during simultaneous downloads and calls even on fast connections.

### Best Practices for Accurate Testing

**Use a wired ethernet connection.** Plug your laptop or desktop directly into your router with an ethernet cable. WiFi introduces its own variability — wireless congestion, distance from router, interference — that's separate from your ISP's actual performance.

**Test at the right time.** Test during peak evening hours (7–10 PM) AND during off-peak hours (morning or midday). The difference shows whether your connection suffers from congestion. ISPs are contractually responsible for their peak-hour performance.

**Close all other applications.** Background programs that download updates (Windows, apps), sync files (Dropbox, iCloud), or stream (other devices) all consume bandwidth during testing.

**Test multiple times.** Run three to five consecutive tests and average the results. Single tests can be outliers.

**Test from multiple servers.** On Speedtest.net, try the auto-selected server AND a server in your ISP's network AND a server in a different city. This reveals routing issues.

**Test different devices.** Your laptop's network card has a maximum throughput. On a gigabit connection, older network cards may cap at 500 Mbps. Use a device capable of handling your connection speed.

### Interpreting Your Results

**Download speed:** Should be close to your plan's advertised speed (within 80–90%). If consistently below 70%, contact your ISP.

**Upload speed:** On fiber plans, should match download closely. On cable, expected to be much lower. If your cable upload is below 10 Mbps on a 500 Mbps download plan, your ISP's equipment may need attention.

**Ping/Latency:** Under 20ms is excellent for most use. 20–50ms is good. Over 100ms is problematic for real-time applications.

**Jitter:** The variation in latency from packet to packet. Low jitter (under 5ms) means consistent, smooth performance. High jitter causes choppy audio and video on calls.

### When You're Not Getting What You're Paying For

If repeated tests during peak hours show significantly below-promised speeds: 1. Reboot your modem and router and test again 2. Check all cable connections from the wall to your modem 3. Test with a different device wired directly to the modem (bypassing your router) 4. Call your ISP with your test results and timestamps — documentation helps

Most ISPs have a service guarantee clause that requires them to deliver a meaningful percentage of advertised speeds.

Use [FiberFinder's address lookup](/availability) to see every provider available at your specific address.