How do you run an accurate internet speed test?

For the most accurate result, we recommend running the test on a device connected to the modem with a network cable (Ethernet) if possible. If you use a wireless connection, it helps to choose the 5 GHz network and stay close to the modem. Pausing the download and video traffic of other devices on the same network, and closing heavy apps open on your test device, makes the result realistic. We recommend running the measurement in a single browser tab, without a VPN, and repeating it a few times. The NETMETRİK engine statistically models your speed over the 15-second test; these conditions improve the result's accuracy.

The most common causes of a low result are the conditions at the moment of testing: Wi-Fi distance and interference, the bottleneck of a 2.4 GHz network, other devices on the same network consuming bandwidth, open apps or screen recording keeping your test device's processor busy, an old device or browser, a VPN, and updates running in the background. Once these factors are addressed, the result usually approaches the committed speed.

To see your line's true capacity, a test done with a network cable (Ethernet) is always the most accurate; a wireless connection reflects its own speed limits onto the result. You can also test if you want to measure your Wi-Fi performance, but in that case keep in mind that the result shows not your internet speed but the current quality of your wireless connection.

The 5 GHz band offers higher speed and less interference but has a shorter range; 2.4 GHz reaches farther but is slower and clashes more with neighboring networks. If you have a high-speed fiber or cable subscription, a test over 2.4 GHz may measure the limit of the Wi-Fi band rather than your line. For the highest result, use the 5 GHz network close to the modem.

Yes. Screen recording, live streaming or video calls simultaneously occupy your device's processor and network connection; this makes the measured speed appear lower than it is. For the most accurate result, run the test with these apps closed.

They do. Games, cloud backups, large downloads or many tabs open on a phone, computer, tablet or TV share both the processor and the connection. The more idle your test device is, the more realistic the result.

On shared networks such as workplaces, schools, hotels and cafés, bandwidth is shared among many users and a firewall, proxy server or traffic prioritization is often applied. For this reason a result on a shared network does not reflect your personal line; to measure your own subscription, a test on your home network gives a more accurate result.

RPM (Round-trips Per Minute) is a responsiveness metric defined by RFC 9097 that measures how responsive your connection stays under load. A high RPM means low latency even when the line is full; this provides smoothness in interactive uses such as video calls, cloud gaming and live streaming. Beyond a single speed figure, NETMETRİK reports RPM as a first-class output.

Bufferbloat is the abnormal swelling of latency when your connection is under load (for example during a large upload). High bufferbloat causes freezing in video calls and lag in games even when your speed is good. The NETMETRİK test compares the idle ping with latency under load and shows bufferbloat as a separate value.

Internet speed is not a single static number but a distribution of measurements that depends on conditions. The NETMETRİK Quality Framework (NQF) models latency stability, responsiveness under load (RPM) and bufferbloat alongside bandwidth. This way you see in full whether your connection is sufficient for experiences like 4K video, cloud gaming or video calls. Our methodology is open and academically published.

Mbps (megabits per second) is the standard unit of internet speed; MB/s (megabytes per second) is what you see in file download speed. Since 1 byte is 8 bits, the MB/s value is roughly one eighth of Mbps. For example, a 100 Mbps connection corresponds in practice to about 12.5 MB/s download speed.

On mobile networks, speed changes instantly depending on location, base station density, cell capacity and signal strength. For this reason a result on mobile data reflects the conditions of that moment and that spot. For a consistent measurement, it is recommended to take several tests in the same place.

NETMETRİK is an independent measurement infrastructure; our measurement engine, measurement servers and main database are in Türkiye. Your data is never sold and is never shared with third parties for marketing. To run the service we work only with infrastructure providers (such as content delivery and encrypted backups) and, when needed, a location service that finds your city from your IP address; section 8 of the Privacy Policy has the details.

First, we recommend repeating the test a few times with a network cable, with other devices off and at different times. If the result is still significantly below your subscription, it helps to restart your modem and check the cable connections. If the situation persists, you can take a screenshot of your result screen and send it to your internet service provider; since the NETMETRİK result page also shows the date and time of the measurement, it gives your provider concrete evidence.