Internet Speed Test
An internet speed test measures three numbers: how fast you can download from the internet, how fast you can upload, and round-trip latency to a test server. Busuk uses the open NDT7 protocol from M-Lab (the same one Google's built-in speed test uses), so the results are reproducible and the measurement code is auditable. The methodology page goes deeper on how NDT7 measures throughput.
Run the test
The widget below runs the NDT7 protocol against your nearest M-Lab server. Click Start Speed Test, give it about 25 seconds, and you will get download, upload, and latency numbers comparable to what Google Fiber's built-in speed test reports. For best accuracy, test on a wired connection close to your router and close other apps using bandwidth in the background.
What the test measures
A complete speed test reports four numbers. Each one tells you something different about your connection:
- Download speed (Mbps). How fast you can pull data from the internet: page loads, video streams, downloads. Headline ISP plans quote download speed first.
- Upload speed (Mbps). How fast you can push data out: cloud backups, video calls, posting media. Most home plans are asymmetric: a 500/50 plan gives 500 Mbps down, 50 Mbps up.
- Latency / ping (ms). Round-trip time for a tiny packet to reach the test server and come back. Anything under 30ms feels instantaneous; over 100ms, video calls start lagging and online gaming gets unplayable.
- Jitter (ms). How much your ping fluctuates between consecutive measurements. Steady 50ms is fine for streaming. Wildly varying 20–80ms is jitter, and it kills voice/video calls even when average latency looks acceptable.
What's a good speed for what you do?
Speed needs depend entirely on what's running on your network:
| Activity | Minimum speed | Latency matters? |
|---|---|---|
| Web browsing & email | 5 Mbps down | No |
| SD video streaming (one stream) | 3–5 Mbps down | No |
| HD video streaming (one stream) | 5–15 Mbps down | No |
| 4K video streaming (one stream) | 25 Mbps down | No |
| Video conferencing (Zoom/Meet) | 1.5–3 Mbps both ways | Yes (under 100ms) |
| Online gaming | 5–25 Mbps | Yes (under 50ms) |
| Cloud backup / large uploads | 10+ Mbps up | No |
| Heavy household (4 users + smart home) | 100+ Mbps down | Mixed |
| Multi-stream 4K + remote work | 300+ Mbps down | Mixed |
For mixed-use households, double the per-activity minimum to leave headroom for background traffic (security cameras, smart-home polling, OS updates).
Why your speed might be below your plan
ISPs quote peak speeds under ideal conditions on a wired connection. Real-world numbers are usually 20–30% lower. If your test result is more than 30% below the plan, something is off. Five common causes:
- You're testing on Wi-Fi. Wi-Fi adds 10–30% overhead even with a perfect signal. Test on Ethernet directly into the router. If Ethernet matches your plan but Wi-Fi doesn't, the issue is your wireless setup, not your ISP.
- Distance from the router. Wi-Fi signal halves roughly every 10 metres of free space, faster through walls. If you're testing from another room, move within 3 metres of the router and retest.
- Peak-hour congestion. Cable and shared-fiber connections slow down between 7pm and 11pm as neighbors stream. Test at 3am. If speeds are full at night, your ISP is over-subscribing the local node.
- Outdated router or firmware. A router rated for 100 Mbps cannot deliver 500 Mbps no matter what your ISP plan says. Check the router brand directory for your model's WAN throughput rating; update firmware via the admin panel.
- ISP-side traffic shaping or throttling. Some ISPs throttle specific protocols (BitTorrent, video streams) or specific times. Run the test from a VPN. If speeds suddenly improve, your ISP is shaping. The ISP-specific guides document known throttling behavior per provider.
How to improve your speed
Once you've identified the bottleneck, the fix depends on where it is:
- Wi-Fi bottleneck: upgrade to Wi-Fi 6 (or Wi-Fi 6E if your devices support it), reposition the router central and elevated, switch to 5 GHz for nearby devices.
- Router bottleneck: log in to the router admin panel and check WAN throughput; if it's capped, upgrade the router.
- ISP plan bottleneck: if Ethernet matches plan but plan is too slow for your usage, upgrade plans. Consult ISP-specific guides for the available tier above yours.
- Modem bottleneck: rare on fiber, common on cable. DOCSIS 3.1 modems handle gigabit; older DOCSIS 3.0 modems cap around 600 Mbps regardless of plan.
Why Busuk uses NDT7
NDT7 is the speed test protocol maintained by Measurement Lab (M-Lab), a research consortium founded in 2009 by Google, the New America Foundation, and academic partners including Princeton University. The full server code, client code, and measurement methodology are open source under the Apache 2.0 license.
That matters because most popular speed tests are proprietary. Speedtest.net (Ookla), fast.com (Netflix), and ISP-branded tests all run measurements through code you cannot inspect, against servers whose locations and configurations you cannot verify. NDT7 publishes everything: the math, the code, the server topology, and the historical results dataset.
For a deeper breakdown of how NDT7 measures throughput and how to interpret edge cases like high-jitter connections, see our methodology page and results guide.
Frequently asked questions
What is a good internet speed?
For one or two users browsing and streaming HD, 25–50 Mbps download is enough. For 4K streaming, 25+ Mbps per stream. For online gaming, raw speed matters less than latency; aim for under 50ms ping. A four-person household with multiple devices generally wants 100+ Mbps to avoid contention.
Why is my speed lower than what my ISP advertises?
ISP plans quote peak speeds achieved under ideal conditions on a wired connection. Real-world speeds are usually 20–30% lower due to Wi-Fi overhead, distance from the router, contention with neighbors on the same node, and ISP traffic shaping during peak hours. Test on a wired client at 9pm and at 3am. If both are below plan by 30%+, file a support ticket.
What is NDT7?
NDT7 is an open speed-test protocol developed by Measurement Lab (M-Lab), a research consortium that includes Google and academic partners. It uses standard TCP and BBR congestion control to measure throughput, making results reproducible and auditable. Unlike proprietary tools like Speedtest.net, the NDT7 server code, client code, and measurement methodology are all open source.
How is upload speed different from download speed?
Download is data flowing into your network (web pages, video streams). Upload is data flowing out (video calls, cloud backups, posting photos). Most home connections are asymmetric. A 500/50 plan gives 500 Mbps download but only 50 Mbps upload. For video conferencing or working from home, upload often matters more than download.
Why does my ping change between tests?
Ping (round-trip latency) varies with network congestion, route changes between you and the test server, and Wi-Fi interference. A few milliseconds of variation between back-to-back tests is normal. Sustained variation of 30+ms is called jitter, and it disrupts video calls and online gaming even when raw speed is fine.
Should I use Wi-Fi or Ethernet for the test?
Ethernet for the cleanest measurement of your ISP connection. Wi-Fi adds variable overhead (signal strength, channel congestion, distance) that masks actual ISP performance. Test on Wi-Fi when you want to measure your home Wi-Fi quality. The difference between Ethernet and Wi-Fi numbers tells you whether your router or your ISP is the bottleneck.