How to Test Internet Speed

1. The 9-step testing workflow

Too many people click a speed test once while a 4K movie streams in the next room, see a lower number than their plan states, and assume their internet provider is broken. An accurate speed test is an experiment that requires controlling variables. Follow this disciplined workflow:

1. Decide what connection you want to evaluate: Are you testing your broadband provider’s incoming line, or your home Wi-Fi coverage?
2. Choose the device that matters: Test on the specific hardware you use for work, streaming, or gaming.
3. Prepare the connection: Minimize competing household downloads and close heavy background applications.
4. Run your baseline test: Execute the real speed test above to capture unburdened throughput and idle latency.
5. Record the complete metric set: Note download Mbps, upload Mbps, ping latency, jitter, and packet loss.
6. Repeat the test: Run two additional tests under identical conditions to confirm consistency.
7. Compare conditions: Compare wired vs. Wi-Fi, or compare morning results against peak evening hours.
8. Interpret the measurements: Compare your numbers with your broadband provider’s tier and the bandwidth requirements of your daily applications.
9. Retest when troubleshooting: If you make an adjustment (like repositioning your router or switching Wi-Fi bands), retest immediately to verify the impact.

2. Step-by-step testing guide

Follow these detailed steps to perform a fair, repeatable internet speed measurement on any computer, tablet, or smartphone:

Step 1: Choose the device that answers your question

If you want to know whether your internet service provider (ISP) delivers the speed you pay for, use a modern desktop or laptop with a wired Ethernet port. If you want to know why a video call stutters in your home office, test on that exact laptop in that exact room.

Step 2: Choose between Wi-Fi and Ethernet

Ethernet isolates your broadband connection from walls, physical distance, and radio interference. Wi-Fi measures the wireless path that most phones and smart devices actually rely on. Neither is "wrong," but they measure different segments of your network.

Step 3: Pause heavy background downloads and uploads

Check other devices in your home. A single active game patch on a console (such as Steam or PlayStation Network) or an automatic cloud photo sync can saturate your connection, leaving only a fraction of your bandwidth for the test.

Step 4: Close background applications and browser tabs

Close file-syncing utilities, peer-to-peer software, and streaming tabs. Temporarily disconnect from virtual private networks (VPNs) and corporate security tunnels unless you specifically intend to evaluate the VPN’s encryption overhead.

Step 5: Run the real speed test

Press the Start Test button at the top of this page. Keep the browser tab in the foreground while the engine completes the latency probe, multi-stage download test, upload measurement, and packet-loss check.

Step 6: Record all five core measurements

Do not look only at download speed. Record all five figures: download (Mbps), upload (Mbps), idle ping (ms), jitter (ms), and packet loss (%). If you expand the Test details accordion, also note your ping under load (loaded latency).

Step 7: Repeat the test three times

Run the test three times in succession, waiting 30 to 60 seconds between runs. If two runs show 480 Mbps and one shows 460 Mbps, you have consistent throughput. If results swing wildly (e.g., 500 Mbps down to 80 Mbps), investigate local Wi-Fi interference or background network bursts.

Step 8: Test at different times of day

If you suspect evening slowdowns, test once in the morning (between 8:00 a.m. and 11:00 a.m.) and once during peak evening hours (between 7:00 p.m. and 11:00 p.m.). This establishes whether neighborhood traffic congestion affects your line.

Step 9: Compare with your broadband plan and application needs

Check your provider's monthly bill or broadband consumer label. Compare your wired baseline against your advertised plan. For everyday context, compare your results with our dedicated guides for streaming requirements and video call bandwidth.

Step 10: Use dedicated troubleshooting if results show a problem

If your wired baseline is significantly lower than your plan, or if your Wi-Fi drops sharply across the hallway, consult how to fix slow internet and how to improve internet speed for targeted fixes.

3. Which device should you test?

No single device is universally the "correct" choice for every test. Your choice of testing hardware depends directly on the problem you want to solve:

• For an ISP broadband baseline: Use a relatively modern desktop or laptop connected via a Gigabit Ethernet cable directly to your router or gateway. This eliminates Wi-Fi attenuation and ensures client processor limitations do not cap your result.
• For daily work and video calls: Test on your work laptop from your designated desk. Testing next to your router will not tell you whether your webcam video will freeze during a meeting in the upstairs bedroom.
• For mobile phone & tablet experience: Test directly on your smartphone in the rooms where you browse or stream social media. This evaluates your phone’s specific internal antenna design and Wi-Fi chip.
• For smart TVs & streaming boxes: If your TV or streaming stick suffers from buffering, run a browser-based speed test on that device or place your phone right next to the TV to test wireless signal strength in that exact physical spot.
• For gaming latency issues: Test on your gaming PC or console. Latency is heavily influenced by client network adapters, background software, and Ethernet routing.

4. Wi-Fi vs. Ethernet testing

Understanding the fundamental differences between wired and wireless testing is the single most important concept in speed testing. Wi-Fi and Ethernet serve two distinct diagnostic purposes:

Important note: Connecting via Ethernet does not magically increase the speed of your internet service plan. It simply removes in-home wireless loss, allowing your device to access the full throughput your router receives from the provider. For dedicated wireless testing, see our guides on Wi-Fi speed testing and how to test Wi-Fi speed.

5. Clean baseline vs. real-world testing

Speed tests should be categorized into two distinct testing types depending on what you want to learn:

The clean baseline test

A clean baseline test measures your connection in an idealized state. You pause all other household devices, close unnecessary background apps, turn off VPNs, and connect with a wired cable. This answers: "What is the maximum capacity my broadband provider and router are capable of delivering right now?" If your baseline test fails to reach your plan's expected speed, the issue lies in your ISP line, modem, or core router hardware.

The real-world test

A real-world test measures what your device experiences under normal, everyday operating conditions. You leave household smart TVs, family phones, and background syncs active, and you test over Wi-Fi from your couch. This answers: "How much bandwidth and latency does this specific device actually get while the rest of my household lives their digital lives?" If your clean baseline is 500 Mbps but your real-world test is 45 Mbps, the problem is not your ISP plan—it is household traffic competition or Wi-Fi signal loss.

6. How many times & when to test

A single speed test is a snapshot of a single 30-second window across a dynamic, constantly shifting global network. It should never be treated as an immutable grade.

The 3-test rule of thumb

Whenever you test, run three consecutive tests about one minute apart. Look at the numbers:

• If all three results are within 10% of each other (e.g., 290 Mbps, 310 Mbps, 305 Mbps), you have established a reliable average.
• If one test is dramatically lower (e.g., 300 Mbps, 70 Mbps, 295 Mbps), that outlier was likely caused by a momentary background download, a sudden Wi-Fi packet retransmission, or a client CPU spike. You can safely discard the outlier.

Accounting for time of day

Internet usage follows predictable societal patterns. Residential networks typically experience peak congestion between 7:00 p.m. and 11:00 p.m. on weekday evenings, when neighborhoods stream high-definition video and download gaming updates simultaneously. If your connection feels sluggish in the evening, test once at 10:00 a.m. and again at 8:30 p.m. Comparing those tests reveals whether local node congestion or neighborhood oversubscription is affecting your performance.

7. Testing in the right physical location

When testing over Wi-Fi, physical location dictates your results. Wi-Fi radio waves attenuate rapidly through building materials like brick, concrete, drywall, metal studs, and tile:

• The control test (near the router): Test 5 to 10 feet away from your router with a clear line of sight. This tells you the highest wireless speed your router and device can negotiate together.
• The operational test (where you use the device): Test in the bedroom, home office, or living room where you normally sit. The gap between your control test and your operational test shows exactly how much speed you lose to walls, floors, and physical distance.

Never stand right next to the router to test your connection and then assume that same speed reaches a smart TV two floors up.

8. Speed test metrics explained

Our speed test engine measures five fundamental connection variables. Each variable reflects a distinct dimension of network quality:

9. What does Mbps actually mean?

Internet speeds are measured in Mbps (megabits per second), whereas file sizes on your computer are measured in MB (megabytes). Confusing these two units is one of the most common user misunderstandings:

• There are 8 bits in 1 byte. Therefore, an internet connection of 8 Mbps can theoretically transfer 1 megabyte of data per second (1 MB/s).
• A 100 Mbps internet connection transfers up to 12.5 MB/s under ideal theoretical conditions.
• A 1,000 Mbps (Gigabit) connection transfers up to 125 MB/s under ideal conditions.

In practice, real-world file transfers will always measure slightly lower than the theoretical conversion due to networking protocol overhead (TCP/IP packet headers, TLS encryption, and packet acknowledgments), which typically consumes 5% to 10% of total bandwidth.

10. Advertised plan speed vs. measured speed

When you subscribe to a "500 Mbps" broadband plan, your service provider is advertising the maximum theoretical data throughput provisioned to your modem or optical network terminal (ONT). Your measured speed test result will often differ from that advertised number for several valid technical reasons:

• Wi-Fi signal degradation: Moving just 20 feet away through two interior walls can cut Wi-Fi throughput by 50% to 70%.
• Client device hardware: An older smartphone with a single Wi-Fi antenna (1x1 SISO) cannot physically process the same speeds as a new laptop with a 2x2 Wi-Fi 6E radio.
• Router CPU limitations: Saturated router processors running stateful packet inspection or older NAT architectures can bottleneck traffic.
• Local network overhead: Other background devices sharing your local connection consume bandwidth concurrently.
• Broadband consumer labels: The Federal Communications Commission (FCC) requires U.S. providers to display typical speeds rather than theoretical maximums on their consumer labels. A wired test showing 85% to 95% of your plan speed indicates an entirely healthy connection.

11. How accurate is a speed test?

No speed test is 100% accurate as a universal measure of the entire global internet. A speed test measures the exact performance observed by your specific device across your local home network, your ISP access network, and the specific routing path to the test server at that exact instant.

Because our speed test runs entirely inside your web browser, its results naturally reflect your browser’s rendering engine, your computer’s background CPU utilization, and your local Wi-Fi environment. This makes the test exceptionally accurate for measuring what your device actually experiences when loading web pages and streaming video, but it does not represent an isolated hardware probe of your provider’s street cable.

12. How this website’s test engine works

We believe in absolute technical transparency. Unlike testing tools that rely on opaque proprietary algorithms, our speed test is built on the open-source @cloudflare/speedtest measurement engine:

• Edge server routing: The engine connects to Cloudflare’s global edge network, automatically selecting the nearest edge location with the lowest latency to your physical location.
• Multi-stage chunk downloads: To measure download speed accurately across all connections—from slow DSL to multi-gigabit fiber—the engine uses dynamic ramp-up chunk sizes. It begins with small 100 KB files to warm up the connection, progressively scaling up through 1 MB, 10 MB, 25 MB, 100 MB, and 250 MB chunks as long as your connection sustains high throughput.
• Upload POST requests: Upload speed is measured by dispatching HTTP POST requests with progressively sized data payloads (100 KB up to 50 MB) to Cloudflare edge endpoints, timing the round-trip completion using the W3C PerformanceResourceTiming browser API.
• Idle vs. loaded latency: The engine measures idle ping first across 20 small packets on an unburdened connection. During the download and upload phases, it continues firing latency probes to measure loaded latency, revealing whether your router suffers from bufferbloat.
• Packet loss via WebRTC TURN: Browsers cannot send raw ICMP ping packets for security reasons. To measure packet loss accurately, our engine utilizes a Cloudflare Realtime TURN server, transmitting 1,000 UDP packets in controlled batches and calculating the exact percentage of dropped or timed-out responses.

13. Why different speed tests disagree

If you run our speed test and then immediately test on another popular platform (such as Ookla Speedtest, Fast.com, or Google Fiber Speed Test), you will often see different results. This does not mean one test is broken or dishonest:

• Different server locations: One test may connect to a server hosted within your ISP’s local datacenter 10 miles away, while another connects to an edge server 80 miles away. Shorter network paths naturally yield lower ping and often higher single-stream throughput.
• Single-connection vs. multi-connection: Some tests use dozens of parallel TCP sockets to artificially saturate a line, while others use realistic chunk requests that mirror real-world browser downloads.
• Different transit and peering: Data travels across different internet service providers, transit exchanges, and peering agreements. A route between your ISP and Cloudflare may have different capacity than a route between your ISP and Netflix.
• Test duration: Brief 5-second tests measure burst speeds (often inflated by ISP "speed boost" algorithms), whereas sustained 20-second tests measure true continuous throughput.

14. Why one test is not the whole internet

It is crucial to remember: a speed test measures the connection between your device and the test server. It cannot guarantee how any specific third-party website, game server, or streaming service will perform:

If you measure 800 Mbps on our speed test but download a game from a remote server at only 40 Mbps, the bottleneck is almost certainly the remote game server’s upload capacity or congested peering along that specific route, not your local internet connection. A speed test tells you what your line is capable of handling, not what every remote web server chooses to feed you.

15. Common speed-testing mistakes

Avoid these ten common testing errors to ensure your results are valid and actionable:

1. Testing while cloud backups are syncing: Background services like iCloud, Google Photos, or OneDrive upload silently, which starves your connection and inflates ping latency.
2. Testing over distant Wi-Fi and blaming the ISP: If your Wi-Fi test in the upstairs bedroom is slow, test over Ethernet before calling your provider to complain.
3. Relying on a single test run: A single test can be skewed by a momentary background update or temporary Wi-Fi collision. Always run at least three tests.
4. Testing on an old, underpowered device: An eight-year-old laptop or budget tablet may bottleneck Gigabit internet simply because its processor or network card cannot handle the throughput.
5. Ignoring upload speed: Many users only check download speed, ignoring upload. Saturated upload links choke outgoing acknowledgments and ruin video calls.
6. Ignoring ping, jitter, and packet loss: A connection with 500 Mbps download but 120 ms ping and 4% packet loss will feel terrible for gaming and Zoom meetings.
7. Leaving a VPN connected: Testing through an encrypted VPN tunnel measures the VPN server’s capacity and route, not your direct broadband connection.
8. Comparing tests taken under completely different conditions: Comparing a morning wired test against an evening Wi-Fi test will lead to false conclusions.
9. Assuming the highest momentary burst is your true speed: Short speed spikes during the first three seconds do not reflect sustained continuous performance.
10. Assuming a low test result proves your ISP is throttling you: In-home Wi-Fi interference, faulty Ethernet cables, and router queueing cause far more slowdowns than intentional ISP throttling.

16. Testing on mobile phones & 4G/5G

When testing on a smartphone, you must clearly distinguish between testing your home Wi-Fi and testing your cellular carrier network:

• To test home Wi-Fi on a phone: Verify that Wi-Fi is enabled and connected to your home network. Stand where you normally use your phone. Ensure your phone connects to the faster 5 GHz band rather than 2.4 GHz if your router broadcasts separate network names.
• To test mobile data (4G LTE or 5G): Open your phone’s settings and turn Wi-Fi completely off so all data is forced through your cellular carrier. Look at your phone’s status bar to confirm whether you have an active 4G or 5G signal. For dedicated cellular benchmarks, visit our mobile speed test, 4G speed test, and 5G speed test pages.

Keep in mind that cellular speeds fluctuate dramatically based on your physical distance from the nearest cell tower, the number of users connected to that sector, and whether you are indoors behind energy-efficient glass or concrete walls.

17. Testing home Wi-Fi performance

To evaluate your home Wi-Fi network systematically, follow this 5-step mapping procedure:

1. Start next to the router: Run a test 5 to 10 feet from your router in clear line of sight. This is your maximum Wi-Fi baseline.
2. Walk to primary living areas: Test in your living room, kitchen, bedroom, and home office. Stand exactly where you sit when using devices.
3. Record the drop-off: Note the difference in Mbps and ping latency between the router room and distant spaces. A drop of more than 50% indicates significant wall attenuation.
4. Identify dead zones: If packet loss appears or ping jumps from 20 ms to over 100 ms, you have found a wireless dead zone or high-interference area.
5. Consult the dedicated Wi-Fi guide: For advanced techniques like testing 2.4 GHz vs. 5 GHz bands and optimizing channel widths, read how to test Wi-Fi speed around your home.

18. Testing when internet feels slow

If your internet connection feels sluggish or keeps buffering, use the speed test as a diagnostic triage tool:

1. Run an immediate baseline test: Capture what the connection is doing right now while the problem is occurring.
2. Test a second device: If your laptop is slow but your phone achieves 300 Mbps in the same room, the issue is isolated to your laptop’s software or network adapter.
3. Compare Wi-Fi with a wired cable: If a wired computer runs at full plan speed while Wi-Fi crawls, your router’s wireless radio or channel is congested.
4. Check latency and packet loss: If download speed looks normal but video calls freeze, look for packet loss (above 1%) or high jitter (above 30 ms).
5. Proceed to troubleshooting: Once you know whether the issue is local or network-wide, follow our step-by-step diagnostic guide on how to fix slow internet.

19. When should you run a speed test?

Knowing when to test is just as important as knowing how to test. Useful scenarios include:

• When setting up new internet service: Test on your activation day over a wired cable to confirm your provider is delivering the speed you purchased.
• Before calling your ISP’s technical support: Providers take complaints much more seriously when you present a documented log of wired test results rather than a vague complaint that "the Wi-Fi feels slow."
• When video streams buffer or calls drop: Test immediately during the incident to catch temporary packet loss or bandwidth dips in real time.
• When rearranging your home or moving your router: Test before and after moving your router to verify whether its new position improves coverage.
• When considering an expensive plan upgrade: Test repeatedly to determine whether your household is truly saturating your current plan or whether poor in-home Wi-Fi is the real culprit.

20. How to record your speed test results

When diagnosing intermittent network problems or preparing evidence for your internet service provider, keep a structured testing log. Record tests across several days under comparable conditions:

Our speed test engine also maintains your recent test history locally in your web browser. You can click Share Result after any test to generate a permanent link to share with family or your provider's support team.

21. Interpreting common result patterns

Once your test completes, analyze your full metric profile against these six common performance patterns:

22. What to do after your test

Now that you have executed a fair, repeatable test and recorded your baseline metrics, proceed to the specialized resources in our network quality cluster:

• If your speeds are far below your plan: Follow our comprehensive troubleshooting guide on how to fix slow internet to determine whether your router, modem, or service provider is responsible.
• If your Wi-Fi is the primary bottleneck: Learn practical zero-cost placement tweaks and band optimizations on how to improve internet speed.
• If you need to verify wireless coverage room by room: Use our dedicated Wi-Fi testing guide to map dead zones.
• If your latency spikes under load: Test router queue management with our interactive bufferbloat test and read what is bufferbloat.
• If you want to plan household bandwidth: See our detailed guides for streaming requirements, video call bandwidth, and multi-device planning.