How to Check Your Internet Speed—and Understand the Results

An internet speed test is useful only when the test conditions are clear. A single number taken on a phone at the far end of the house cannot tell you whether your internet plan, modem, router, Wi-Fi coverage, device, or the test server is the limiting factor. The practical goal is to create a repeatable baseline and then change one variable at a time.

This guide explains how to measure a home connection, interpret the four numbers that matter, and isolate common bottlenecks. It is written for ordinary broadband and fiber connections rather than a controlled lab. For more plain-language network guidance, visit the Networking section or return to NexalooTech Home.

Before you test: define the question

Start by deciding what you want to learn. If you want to check whether the provider is delivering close to the subscribed rate, test a computer connected to the main router with Ethernet. If you want to assess real-world Wi-Fi, test the device in the room where it is normally used. Those are different measurements, and both can be valid.

Write down the plan’s advertised download and upload rates. Note whether it is cable, fiber, fixed wireless, DSL, or satellite, because the expected latency and upload behavior vary. Pause cloud backups, game downloads, operating-system updates, streaming, and large file transfers on other devices. A busy household shares the same connection, so background traffic can make a healthy line look slow.

Use a wired baseline first

Connect a capable laptop or desktop directly to the router or gateway with a known-good Ethernet cable. Confirm that the network adapter negotiated at a speed above the internet plan. A gigabit Ethernet link cannot demonstrate a multi-gigabit service, and an older 100Mbps adapter will cap an otherwise faster connection. Disable Wi-Fi on the test computer so the route is unambiguous.

Restarting equipment before every test can hide intermittent faults, so take an initial measurement first. Reboot only if you are deliberately testing whether a fresh connection changes the result. Let the gateway fully reconnect before repeating the measurement.

Run a repeatable three-test sequence

1. Pick a nearby server. Automatic selection is usually sensible, but record the server name. Distance and congestion can alter the result.
2. Run three tests. Leave a minute between them and use the median result rather than choosing the highest number.
3. Repeat at another time. Compare morning, evening, and any period when the connection normally feels slow.
4. Keep the setup unchanged. Use the same device, cable, server, and browser or app for comparisons.

A browser test is convenient, while a reputable native app can reduce browser overhead on very fast lines. Router-based tests are valuable because they measure from the gateway and remove the client device and Wi-Fi from the path. They may still use a different server or method, so compare trends rather than treating unlike tools as interchangeable.

What the results actually mean

Download speed describes how quickly data reaches you. It affects large downloads, streaming headroom, and how fast cloud files arrive. Upload speed matters for video calls, sending media, remote backups, and synchronizing shared folders. Cable plans are often asymmetric, while many fiber plans are closer to symmetrical.

Latency, usually shown in milliseconds, is the round-trip delay. It shapes the responsiveness of calls, games, remote desktops, and ordinary browsing even when raw throughput is high. Jitter is variation in that delay. A connection with adequate bandwidth but unstable latency can produce choppy audio or erratic game play. Packet loss, when reported, means some data did not complete the trip and deserves investigation.

No public speed test represents every destination on the internet. The result is a snapshot between your device and a particular test server. A remote service can be slow even when the local access line is healthy, while a nearby test server may show excellent throughput that a distant service cannot match.

Separate an internet problem from a Wi-Fi problem

After establishing the wired baseline, run the same test over Wi-Fi beside the router. Then move to the room where performance is poor. A strong wired result with a weak nearby Wi-Fi result points toward router settings, radio congestion, or the client. A good nearby result that collapses with distance points toward coverage, obstacles, or interference.

Use the 5GHz or 6GHz band for higher speeds at shorter range and 2.4GHz for longer reach when appropriate. Do not combine results from different bands without recording the difference. Thick masonry, metal, mirrors, appliances, neighboring networks, and poor router placement can all matter. Put the router in an open, elevated, central position where practical.

Smart devices typically need little bandwidth, but they depend on stable coverage and correct setup. If a connected thermostat frequently drops offline, the diagnostic approach in our smart thermostat buying guide helps you evaluate Wi-Fi and compatibility before replacing hardware.

Read common result patterns

Wired and Wi-Fi are slow at all times

Check the Ethernet link rate, modem signal indicators, cables, and the provider’s status information. Test directly through the main gateway rather than a secondary mesh node. If multiple wired devices show the same repeatable shortfall, document dates, servers, and screenshots before contacting the provider.

Speeds fall mainly in the evening

Repeat the wired test across several days. A consistent peak-time decline can indicate shared-network congestion, but first rule out scheduled backups and household activity. Cloud synchronization can consume upload capacity and increase latency for everyone. Our guide to cloud storage and file sharing services explains why selective sync and bandwidth controls matter.

Download is fine, but calls break up

Look at upload speed, loaded latency, jitter, and packet loss. Saturating a modest upload channel can make interactive traffic wait. Pause outgoing backups, camera uploads, and peer-to-peer transfers, then retest. Router quality-of-service settings may help when correctly configured, but they cannot create capacity the line does not have.

Only one device is slow

Update its operating system and network driver, forget and rejoin the Wi-Fi network, compare it with another device in the same position, and check power-saving settings. Browser extensions, security software, a VPN, or an overloaded CPU can also reduce measured throughput. Test without a VPN when diagnosing the access connection, then test again with it to understand the privacy-performance tradeoff.

How much speed do you need?

There is no honest universal number. Count simultaneous activities, not just people. Several video streams, a large game update, a cloud backup, and two calls can overlap. Leave headroom instead of planning around the exact theoretical sum. Upload capacity deserves equal attention for remote workers, creators, and households with security cameras.

For business use, consistency, support response, failover options, and service terms can be more important than a headline maximum. The computers also need interfaces that support the network you buy; our business laptop guide covers Wi-Fi, docking, ports, manageability, and other fleet considerations.

A practical troubleshooting record

Create a small table with date, time, test device, wired or wireless connection, room, band, server, download, upload, latency, jitter, and packet loss. Add a note about active household traffic. Five disciplined measurements are more useful to a support technician than dozens of unexplained screenshots.

If the wired baseline repeatedly misses expectations, replace questionable cables, bypass optional switches, and retest. If only Wi-Fi is affected, adjust placement and channels before buying new equipment. If one service alone is slow, compare another destination and check that service’s status. This process turns “the internet is slow” into a specific, testable problem—and that is what makes a speed test genuinely useful.