To check your internet speed, open a browser and visit Cloudflare Speed Test or another trusted testing service. Start the test and read your download speed, upload speed, latency (ping) and, where provided, jitter. For a reliable result, pause heavy downloads and repeat the test several times.
A speed test measures how your current device connects to a particular test server at that moment. It does not directly prove the speed of every website, every device or every hour of the day. Compare a Wi-Fi test with an Ethernet test if you need to distinguish a weak wireless connection from your internet service.
Did you know a high download number can coexist with laggy video calls? Bandwidth shows how much data can travel, while latency and jitter describe responsiveness and consistency. Understanding the difference can save hours of changing the wrong setting.
- Quick check: open speed.cloudflare.com and run the browser test.
- Read: download and upload in Mbps; ping and jitter in milliseconds.
- Prepare: pause streaming, uploads and large downloads.
- Repeat: test more than once and at more than one time of day.
- Compare: Wi-Fi near the router, Wi-Fi far away and Ethernet if available.
- Slow result? identify whether the router, device, Wi-Fi or ISP connection is responsible.
- Important: a speed test uses real network data; consider mobile data limits.
In this guide
- How to run an internet speed test
- What download speed means
- What upload speed means
- Mbps versus MB/s: the measurement people often misread
- What ping and latency mean
- What jitter means
- What packet loss means
- Idle latency versus loaded latency
- How to test internet speed on iPhone and Android
- How to test internet speed on Windows or Mac
- Why Wi-Fi speed is lower than the advertised internet plan
- Test with Ethernet to isolate the problem
- Why two speed test websites show different numbers
- How fast is fast enough for streaming?
- Internet speed for gaming and video calls
- Why internet speed changes throughout the day
- What to do if your internet speed is slow
- How to make a fair comparison before contacting your ISP
- A worked example: Wi-Fi slow in one bedroom
- A worked example: high speed but poor Zoom calls
- Privacy and data allowance during speed tests
- Official speed testing references
How to run an internet speed test
Choose a trusted browser-based tool such as Cloudflare Speed Test. Open it on the device and connection you are actually investigating. If you are troubleshooting home Wi-Fi, leave the device connected to that network rather than automatically switching to mobile data.
Stop or pause large file transfers, cloud backups, game downloads and heavy streaming while taking a baseline measurement. Start the test and allow it to measure the relevant transfer directions and response times. Different services may have different button labels and test durations.
Record the date, time, connection type, device and result. A single number has little diagnostic value until you compare it with another test under similar conditions.
- Confirm the correct network is in use.
- Close heavy background data activity where practical.
- Run a reputable speed test.
- Note download, upload, ping and jitter if shown.
- Repeat the test several times.
- Compare another location, device or connection.
What download speed means
Download speed is the rate at which your device receives data from a server. It affects large file downloads, software updates, web page assets and most video streaming. Speed-test services normally express this as megabits per second (Mbps).
A greater number usually means a higher maximum transfer rate during the test. However, the actual time to open a site also depends on server response, congestion, latency and the design of the webpage. A fast connection cannot make a poorly responding server instantaneous.
When comparing results, use the same units. Mbps is not the same as MB/s, which stands for megabytes per second.
What upload speed means
Upload speed is the rate at which your device sends data to the internet. It matters for sending large attachments, uploading video, cloud backup, live streaming and the outgoing video or audio side of a call.
Many consumer plans have different upload and download rates. A download-heavy plan can perform well when watching a movie but be frustrating when sending a multi-gigabyte video. Check what your internet service provider actually promises for each direction.
For an upload complaint, run a test without a simultaneous cloud backup if possible, then test again while normal applications are active. The difference can reveal contention.
Mbps versus MB/s: the measurement people often misread
Internet speed is usually advertised in megabits per second (Mbps). A file manager often reports downloaded data in megabytes per second (MB/s). One byte contains eight bits. Therefore, a theoretical 100 Mbps link corresponds to up to 12.5 MB/s before protocol overhead and practical limitations.
Similarly, a connection measured at 80 Mbps has a theoretical upper data rate of 10 MB/s. Real file transfers can be slower due to server limits, network overhead, wireless conditions and competing users.
Do not compare the number 100 Mbps to 100 MB/s and conclude that the ISP delivered one eighth of the advertised service. Convert the units first.
- 100 Mbps ÷ 8 = 12.5 MB/s, theoretical before overhead.
- 80 Mbps ÷ 8 = 10 MB/s, theoretical before overhead.
- MB/s multiplied by 8 gives approximately Mbps.
- Mbps describes a transfer rate, not a monthly data allowance.
What ping and latency mean
Latency is the delay before data can travel between two endpoints and produce a response. Speed tests often describe a related measurement as ping, usually in milliseconds (ms). For interactive gaming, calls or remote controls, responsiveness may be more important than raw download throughput.
A low ping to a nearby test server does not guarantee a low ping to a distant game server. Geography, routing and the destination network matter. When diagnosing a particular service, measure to a relevant endpoint if the application provides suitable diagnostics.
Cloudflare’s testing information distinguishes unloaded latency from measurements taken during data transfer. A connection may respond quickly when idle but much more slowly while someone is uploading a large file.
What jitter means
Jitter is variation in latency between successive measurements. A connection can have a reasonable average ping but uneven response times. That inconsistency may be noticeable as choppy voices, frozen video or unpredictable game movement.
Cloudflare explains jitter in terms of differences between latency measurements. A smaller, steadier variation is generally preferable for real-time activity, but the exact threshold depends on the application.
Do not interpret jitter alone. Consider upload and download performance, latency under load and packet loss where available. A network-quality problem is usually best diagnosed using several metrics together.
What packet loss means
Packet loss means some units of network data do not arrive successfully. Applications may retransmit data or adapt to missing information, but real-time audio and gaming can suffer interruptions. Some speed tests provide a packet-loss result; not every browser tool does.
Packet loss can result from local Wi-Fi interference, overloaded equipment, congestion or problems along the wider network path. An occasional lost packet on one test is not proof of a specific faulty device.
Use repeated measurements and compare Wi-Fi with a wired connection. If persistent loss appears across devices and a direct wired test, collect evidence for your service provider.
Idle latency versus loaded latency
Many internet connections seem fine when no one is transferring large files but become unresponsive during uploads or downloads. Loaded latency measures how response times change while the connection is busy. A large increase can explain the experience of a call freezing whenever another person starts a big upload.
Cloudflare discusses unloaded, download-loaded and upload-loaded latency in its network testing documentation. Read the context around any latency score rather than comparing one isolated figure with another service’s differently measured result.
If latency under load is the real problem, buying a plan with a higher download headline may not fix your local router, congestion or traffic-management issues. Diagnose first.
How to test internet speed on iPhone and Android
Connect to the desired Wi-Fi network, open a browser and visit a trusted speed test. Make sure mobile data has not taken over because the Wi-Fi signal is poor; check the active connection indicator. Run the test and note the results.
To compare with cellular service, intentionally disable Wi-Fi and run a separate test on mobile data, considering your data allowance. Do not treat the two tests as interchangeable: they use different networks and have different limitations.
Phone performance can be affected by background updates, power conditions, network congestion, the position of the device and the radio capabilities of that model.
How to test internet speed on Windows or Mac
Open a browser on your computer and use the same testing site for repeatable comparisons. If possible, connect directly to the router or network equipment using a suitable Ethernet cable for a second measurement.
The wired test helps isolate the wireless portion of the connection. A high Ethernet result alongside a low Wi-Fi result points toward Wi-Fi coverage, interference, device limitations or radio configuration rather than proving that the ISP’s incoming connection is slow.
Check that your network adapter, cable and router port support the speed you are testing. An older 100 Mbps port cannot demonstrate the full capability of a much faster plan.
Why Wi-Fi speed is lower than the advertised internet plan
An internet plan describes a service tier under stated conditions, while a Wi-Fi speed test measures the end-to-end performance of one device through wireless radio and the wider network. The two are not guaranteed to match. Distance, walls, interference, other users and router design all matter.
A nearby device on a modern router may achieve a different rate from an older laptop two rooms away. A 2.4 GHz connection can have better reach in many homes but more congestion, while 5 GHz or 6 GHz networks may support greater throughput over suitable distances and compatible devices.
Test near the router, then in the location where performance is poor. If the difference is substantial, focus on placement and wireless coverage before changing your internet contract.
Test with Ethernet to isolate the problem
A wired Ethernet test can provide a useful reference for the connection delivered to your router, provided the computer, cable, port and router are capable of the subscribed service tier. It also avoids radio interference during the measurement.
Run the test first on Ethernet with other heavy traffic paused, then on Wi-Fi using the same site and similar time window. If Ethernet is strong and Wi-Fi weak, local wireless conditions are a plausible cause. If both are weak, investigate router, device, test server, network load and ISP conditions.
Do not claim that one Ethernet test certifies the provider is meeting a contractual promise. It is a practical diagnostic comparison, not a full independent service audit.
Why two speed test websites show different numbers
Different speed-test services select different servers, measurement durations and transfer methods. Some use parallel connections; others focus on streaming experience or response times. The route to the test server can also change the measurement.
Cloudflare’s speed-test methodology explains its use of data transfers and latency measurements. Netflix provides Fast.com and its help pages describe the connection-speed recommendations for its own streaming service.
Two different results are not automatically evidence that one tool is dishonest. For troubleshooting, use a consistent test site and record the variation; try a second reputable tool when results seem unusual.
How fast is fast enough for streaming?
The right speed depends on resolution, simultaneous viewers and whether the connection stays stable. As a service-specific example, Netflix currently recommends at least 3 Mbps for 720p HD, 5 Mbps for 1080p and 15 Mbps for 4K UHD for one stream, alongside a stable connection.
Those figures are Netflix’s recommendations, not guarantees for all services or every household. Two 4K streams, a video call and a cloud backup occurring at once create a different demand from one person watching a show.
Remember that a stable connection and low congestion can feel better than a fluctuating line with a larger headline speed. Streaming quality may also depend on the device and subscription tier.
Internet speed for gaming and video calls
Interactive activities need timely delivery, not just high-volume delivery. A connection with hundreds of Mbps can still suffer from latency spikes, jitter or packet loss. For a video call, the outgoing upload path matters because you are sending your camera and microphone data.
Test during the period when the problem usually occurs. If the connection only fails in the evening, an early-morning test will not capture the relevant conditions. Repeat while other household users are active and while they are not.
If calls freeze when a large upload begins, investigate upload saturation and loaded latency. If only one video meeting provider is affected, the issue may not be your general internet speed.
Why internet speed changes throughout the day
Network conditions vary with household usage, local Wi-Fi interference, cellular network load, your provider’s capacity and the server being tested. One peak result at midnight does not describe every evening.
For useful evidence, run comparable tests at the same locations across several days and record the time. Note what other devices were doing. A small set of repeatable measurements is better than one dramatic screenshot.
If the problem is intermittent, write down the symptoms as well as speed results. A call dropping out for five seconds can matter even when the average download speed looks excellent.
What to do if your internet speed is slow
First check whether the slowness affects one website, one device or the entire home. If only one site is slow, the problem may be on that site’s servers. If only one room is slow, inspect Wi-Fi coverage. If every device on Ethernet is slow, examine the modem, router and provider connection.
Restarting network equipment according to the manufacturer’s instructions can resolve temporary problems, but it should not be the only troubleshooting method. Do not factory-reset the router without knowing how to restore configuration and access details.
Make one change, retest and note the difference. This helps identify the actual cause rather than changing several settings and losing the ability to tell what worked.
- Test a second reputable website.
- Test another device on the same network.
- Move nearer to the router.
- Compare Ethernet with Wi-Fi if available.
- Pause large uploads and downloads.
- Inspect router warnings and firmware guidance.
- Contact the provider with repeated measurements if the issue persists.
How to make a fair comparison before contacting your ISP
Keep a simple record: date, time, device, test service, Wi-Fi or Ethernet, measured download and upload, ping, and the observed problem. If the problem is a call, note whether sound broke up or video froze. The symptom matters as much as the number.
A support team can use a series of comparable readings to identify patterns. Avoid reporting a mobile-data speed test as evidence about a home fibre service or an old Wi-Fi adapter as evidence of a gigabit router’s full capability.
If useful, attach a screenshot showing the test result. For a desktop guide, see How to Take a Screenshot on Windows. Redact IP addresses or account data if the recipient does not need them.
A worked example: Wi-Fi slow in one bedroom
Imagine a home with a fast fibre plan. Beside the router, a laptop records a high download result. In a far bedroom, the same laptop tests much slower and video calls pause. Ethernet testing at the router remains strong.
This pattern suggests a local wireless coverage or interference issue is worth investigating. Try a better router location, a different band where compatible or an appropriate mesh/access-point solution. Avoid buying a faster ISP plan until you understand the wireless bottleneck.
Re-test using the same device and service at each location. The improvement should be visible as better speed and more stable calls, not just a stronger signal icon.
A worked example: high speed but poor Zoom calls
Suppose a household sees a 400 Mbps download result but calls freeze whenever someone uploads a large file. The key clue is that the problem occurs under load. Check upload performance and loaded latency rather than focusing only on download speed.
Pause the large upload and repeat the call. If it improves, investigate router traffic management and the available upstream capacity using official equipment or ISP guidance. Do not disable security features as a first response.
The lesson is to match the measurement with the user experience. Download speed alone cannot diagnose every problem.
Privacy and data allowance during speed tests
A real speed test downloads and uploads test data. If you use mobile data or a metered connection, the test may consume some allowance. The amount varies with the service and speed, so avoid repeated full tests when your plan is tightly capped.
Only use testing services you trust, and check their privacy information if you plan to share an account or test report. A result page may reveal network details such as an IP address or approximate provider information.
Do not publish screenshots that show private network identifiers, precise locations, account details or access credentials unless the purpose genuinely requires them.
Official speed testing references
Cloudflare offers a browser test at speed.cloudflare.com, describes its measurements at About the Speed Test and discusses network diagnostics. Its technical guide explains download, upload, unloaded latency, loaded latency and jitter.
Netflix’s recommended internet speeds provides a practical example of how an application translates connection speed into streaming quality. Test services and recommendations can change; consult the official pages for current details.
Frequently asked questions
How can I check my internet speed for free?
Open a browser and use a trusted speed-test service, such as speed.cloudflare.com or fast.com. Record the download and upload results and repeat under similar conditions.
What is a good internet speed?
It depends on what you do and how many people share the connection. Video streaming, large uploads, gaming and calls have different needs. Look at latency and reliability, not only download Mbps.
What is Mbps?
Megabits per second, a unit of data transfer rate. It is not the same as MB/s, or megabytes per second. Divide Mbps by eight for the ideal megabytes-per-second rate before overhead.
What does upload speed mean?
The rate at which your device sends data to the internet. It matters for backing up files, live streaming, sending attachments and outgoing video or audio.
What does ping mean?
A measure of latency or response delay, commonly expressed in milliseconds. Low and consistent latency tends to support responsive games, calls and remote interactions.
What does jitter mean?
Variation in latency across measurements. High variation can cause inconsistent real-time performance even if average ping seems reasonable.
Why is Wi-Fi slower than my internet plan?
The device, Wi-Fi radio, interference, distance, walls, router and network traffic can limit the measured result. Compare a nearby wireless test with a suitable wired Ethernet test.
Should I test over Wi-Fi or Ethernet?
Test Wi-Fi to measure your actual wireless experience. Test Ethernet with compatible equipment to help isolate the incoming internet connection from local wireless limitations.
Why do speed-test websites disagree?
They may use different servers, routes and testing methods. Compare repeated results using the same service and conditions before concluding there is a problem.
Why is my upload speed slower than my download?
Many plans offer asymmetric speeds. Check the subscribed service tier and confirm the result with repeated measurements while other uploads are paused.
Can a speed test use my mobile data?
Yes. It transfers real test data and may use your allowance. Verify whether your phone is on Wi-Fi or cellular before running a large or repeated test.
Does a speed test tell me the speed of every website?
No. It measures performance to particular test infrastructure at a particular time. Other websites or games can have different routes, server loads and response times.
Why is my ping high even with fast download?
Bandwidth and latency are different. Distance, routing, congestion, Wi-Fi conditions and loaded traffic can increase delays even on a high-throughput plan.
How many speed tests should I run?
Run several comparable tests, including during the period when you notice a problem. A repeated pattern is more informative than a single peak or low value.
How can I prove slow speed to my ISP?
Keep dated test results, connection type, device details, test service, symptoms and a comparison with Ethernet if available. The provider can advise which diagnostic tests are relevant.
Is 100 Mbps equal to 100 MB per second?
No. 100 Mbps is theoretically equivalent to 12.5 MB/s before real network overhead. Mixing bits and bytes creates misleading comparisons.
Is my router the cause of slow speed?
Possibly, but a slow result may also involve the device, Wi-Fi coverage, a test server or the ISP. Compare wired and wireless devices under similar conditions before blaming a particular component.
The useful takeaway
Test the connection you actually use, read download, upload, ping and jitter together, and compare results under controlled conditions. The most valuable speed test is not the biggest number—it is the one that helps explain why your real activity succeeds or struggles.
Last reviewed: November 2026. Test methodologies, streaming recommendations and ISP offers may change.