How to Test Your Internet Speed Accurately [2026]

Internet Providers in your area

Understanding Speed Test Results

A speed test measures four key metrics. Understanding each helps you evaluate whether your internet is performing as expected:

• Download Speed (Mbps): How fast data travels from the internet to your device. This affects streaming quality, web browsing, and file downloads. Most providers advertise plans based on download speed.
• Upload Speed (Mbps): How fast data travels from your device to the internet. Critical for video calls, uploading files, cloud backups, and live streaming. Fiber typically offers symmetrical upload speeds, while cable and DSL upload speeds are much lower than download.
• Latency/Ping (ms): The round-trip time for data to travel to a server and back. Lower is better. Under 20ms is excellent for gaming and video calls. Under 50ms is good for most activities. Over 100ms causes noticeable delay.
• Jitter (ms): The variation in latency over time. Lower jitter means a more consistent connection. High jitter causes buffering during video calls and lag spikes in games.

Expected Speeds by Provider in your area

Fiber Expected Speeds

Fiber advertises speeds up to 300-5000 Mbps on its 300-5000 Mbps network. Real-world speeds typically reach 80-95% of advertised maximums during off-peak hours. During evening peak times (7-11 PM), expect 70-90% of the advertised speed. Cable connections may see more variation during peak hours due to shared neighborhood bandwidth.

Check speeds in your area: Google Fiber: 1-866-777-7550

Cable Expected Speeds

Cable advertises speeds up to 10-35 Mbps on its 100-1200 Mbps network. Real-world speeds typically reach 80-95% of advertised maximums during off-peak hours. During evening peak times (7-11 PM), expect 70-90% of the advertised speed. Cable connections may see more variation during peak hours due to shared neighborhood bandwidth.

5G Home Expected Speeds

5G Home advertises speeds up to 20-50 Mbps on its 72-300 Mbps network. Real-world speeds typically reach 80-95% of advertised maximums during off-peak hours. During evening peak times (7-11 PM), expect 70-90% of the advertised speed. Cable connections may see more variation during peak hours due to shared neighborhood bandwidth.

DSL Expected Speeds

DSL advertises speeds up to 1-10 Mbps on its 10-100 Mbps network. Real-world speeds typically reach 80-95% of advertised maximums during off-peak hours. During evening peak times (7-11 PM), expect 70-90% of the advertised speed. Cable connections may see more variation during peak hours due to shared neighborhood bandwidth.

Satellite Expected Speeds

Satellite advertises speeds up to 10-20 Mbps on its 25-200 Mbps network. Real-world speeds typically reach 80-95% of advertised maximums during off-peak hours. During evening peak times (7-11 PM), expect 70-90% of the advertised speed. Cable connections may see more variation during peak hours due to shared neighborhood bandwidth.

How to Run an Accurate Speed Test

Follow these steps for the most reliable results:

1. Connect via Ethernet: Plug your computer directly into your modem or router with an Ethernet cable. WiFi introduces variability that makes results less reliable.
2. Close other applications: Shut down streaming services, cloud backups, downloads, and other bandwidth-intensive applications on all devices in your household.
3. Restart your modem and router: Power cycle both devices and wait 2 minutes before testing. This clears any temporary issues affecting performance.
4. Run multiple tests: Test at least 3 times at different hours (morning, afternoon, evening) to get a representative average. Peak hours (7-11 PM) typically show lower speeds.
5. Use a reputable tool: Speedtest by Ookla, Fast.com (Netflix), or your ISP's built-in speed test are all reliable options. Using multiple tools provides the most accurate picture.
6. Check your plan speed: Compare results to your subscribed plan speed, not the provider's maximum advertised speed. Your plan speed is listed on your monthly bill.

Why Your Speed Test Results May Differ

Several factors cause speed test results to differ from your plan speed:

• WiFi vs wired: WiFi connections lose 20-50% of speed due to signal interference, distance from router, walls, and competing devices. Always test with Ethernet for accurate results.
• Time of day: Cable and DSL networks experience congestion during peak evening hours when many users in your area are simultaneously streaming and browsing.
• Network equipment: Older modems and routers may not support your plan's full speed. A DOCSIS 3.0 modem caps out at about 343 Mbps, while DOCSIS 3.1 supports gigabit speeds.
• Server distance: Speed test results vary based on the test server's location. Choose a server in your city or state for the most relevant results.
• Device limitations: Older computers and phones may have network adapters that cannot handle gigabit speeds. Verify your device supports your plan's speed tier.

Speed Requirements by Activity

What to Do If Your Speeds Are Slow

If your speed test results consistently fall below 70% of your plan speed, try these troubleshooting steps in order:

1. Restart your modem and router (power off for 30 seconds, then power on)
2. Test with a direct Ethernet connection to isolate WiFi issues
3. Check for firmware updates on your modem and router
4. Verify no other devices are consuming bandwidth (updates, backups, streaming)
5. Try a different Ethernet cable and different port on your modem
6. If speeds are still low, contact your provider's technical support with your test results, test times, and connection method. Request a line test to check for issues between your home and the provider's network.

Step-by-Step Testing Process

Follow this systematic approach for reliable speed test results that accurately reflect your internet connection's performance. Begin by restarting your modem and router five minutes before testing since this clears any cached connections and temporary issues that may artificially lower your results. Plug your computer directly into the modem via Ethernet cable to bypass your router entirely for the purest measurement of the speed entering your home.

Navigate to speedtest.net or another trusted speed test tool and select a server geographically close to your location for the most relevant results. Distant servers introduce additional latency and potential routing inefficiencies that reduce measured speeds. Run the test three times and record each result including download speed, upload speed, and ping. If any single result deviates more than 20% from the others, run two additional tests and discard the outlier since it likely reflects a temporary network event rather than your typical connection performance.

Why Your Speed Test Results Vary: The Complete Explanation

One of the most common frustrations with internet speed testing is inconsistent results. You run a test and get 450 Mbps, run it again two minutes later and get 280 Mbps, then try a different speed test site and get 520 Mbps. These variations are not random — they result from specific, understandable factors that you can account for once you know what causes them.

Server Distance and Selection

Every speed test works by transferring data between your device and a test server. The physical and network distance to that server directly impacts your result. A speed test server in your city routes through fewer network hops and experiences less latency than a server across the country. Most speed test tools (Speedtest by Ookla, Fast.com, Google's built-in test) automatically select a nearby server, but their definitions of "nearby" differ. Ookla typically selects the lowest-latency server from its network, which may be hosted by your own ISP — potentially giving inflated results since the traffic never leaves your provider's network. Fast.com uses Netflix's CDN servers, which test real-world streaming performance but may not reflect your connection's peak capability.

For the most accurate picture, run tests using at least two different speed test tools and compare results. If Ookla consistently shows 500 Mbps but Fast.com shows 300 Mbps, the real-world performance you experience with most services is likely closer to 300 Mbps, since most internet traffic crosses multiple network boundaries just like Fast.com's test does.

Network Congestion: Time-of-Day Effects

Internet speeds fluctuate throughout the day based on how many people in your area are using the network simultaneously. Cable internet is most susceptible to this: your cable connection shares bandwidth with other homes on the same node (typically 100-500 homes). During peak usage hours — generally 7-11 PM local time when households are streaming, gaming, and downloading — cable speeds can drop 20-40% from off-peak performance. Fiber internet experiences minimal congestion effects because fiber networks have vastly more capacity per user. DSL connections are theoretically immune to neighborhood congestion (each line is dedicated) but can experience congestion at the DSLAM if it is oversubscribed.

To understand your connection's true performance range, run speed tests at three different times: early morning (6-8 AM), midday (12-2 PM), and evening peak (8-10 PM). Record the results for each. The evening result represents your realistic "worst case" performance — this is the speed you can rely on during the hours when you are most likely using the internet. If evening speeds are dramatically lower than morning speeds (more than 30% reduction), your provider may be oversubscribing your neighborhood node, and you should contact them about the issue.

Wi-Fi vs. Ethernet: The Hidden Bottleneck

The single largest source of inaccurate speed test results is testing over Wi-Fi instead of a wired Ethernet connection. Wi-Fi introduces multiple performance-reducing factors that have nothing to do with your internet connection speed: signal interference from walls, appliances, and neighboring Wi-Fi networks; protocol overhead from encryption and error correction; bandwidth sharing between all Wi-Fi devices on the same channel; and physical distance from the router.

A concrete example: a household with a 1 Gbps fiber connection and a Wi-Fi 6 router might see 600-700 Mbps on a speed test from a device in the same room as the router, 300-400 Mbps from one room away, and 50-150 Mbps from the far side of the house. None of these results reflect the actual internet connection speed — they all reflect Wi-Fi limitations. Only a speed test run over Ethernet from a device connected directly to the router (or directly to the ONT/modem) shows your true internet speed. If you do not have an Ethernet-capable device, test from a Wi-Fi device positioned within 5 feet of the router with clear line-of-sight for the most representative Wi-Fi result.

Step-by-Step: The Most Accurate Speed Test Method

Follow this protocol for speed test results that accurately represent your internet connection's capability.

Preparation (5 Minutes)

1. Connect via Ethernet. Use a Cat5e or Cat6 cable to connect your computer directly to your router or modem. If your computer lacks an Ethernet port, use a USB-to-Ethernet adapter (available for $10-15 and well worth owning for troubleshooting).
2. Close all other applications. Cloud syncing (Dropbox, Google Drive, iCloud), streaming services, software updates, and browser tabs with auto-refreshing content all consume bandwidth during your test. Close everything except the browser tab you will use for the speed test.
3. Disconnect or pause other devices. If possible, temporarily pause or disconnect other devices from your network. Smart TVs, gaming consoles, security cameras, and other household devices generate constant background traffic. On most routers, you can temporarily enable a "guest network" and connect only your test device to isolate it from other traffic.
4. Verify no VPN is active. VPN connections route your traffic through an additional server, adding latency and potentially limiting throughput. Ensure any VPN software is fully disconnected before testing.

Execution (10 Minutes)

1. Run Ookla Speedtest (speedtest.net). Allow the tool to auto-select the server. Record download speed, upload speed, and ping (latency). Run the test three times with 30-second pauses between tests.
2. Run Fast.com. This Netflix-powered test automatically starts. Wait for it to complete, then click "Show more info" to see upload speed and latency. Record all three values.
3. Run your ISP's speed test (if available). Many providers (Xfinity, AT&T, Verizon) offer their own speed test tools. These test within the provider's network and typically show the best possible results. While potentially inflated for real-world usage, they are useful for comparison.
4. Run Google's speed test. Search "internet speed test" in Google and click "Run Speed Test." Google's test uses Measurement Lab (M-Lab) servers, which provide an independent third-party measurement. M-Lab results tend to be more conservative but highly reliable.

Interpretation

Average the results from each test tool separately, then look at the overall range. Your real-world internet performance falls somewhere between the lowest and highest averages. If all four test tools show similar results (within 15% of each other), you have high confidence in your connection's actual speed. If results vary widely (more than 30% difference between the highest and lowest tool), network routing or server location effects are significant, and you should weight the Fast.com and M-Lab results more heavily as they better represent real-world multi-hop performance.

What Your Speed Test Results Mean: Practical Interpretation

Download Speed Benchmarks

• Under 25 Mbps: Below the FCC's previous broadband threshold. Adequate for single-user email and web browsing. One HD stream is possible but may buffer during peak hours. Multiple simultaneous users will experience frustration. Consider upgrading if alternatives are available.
• 25-100 Mbps: Suitable for small households (1-3 people) with moderate usage. Supports 2-3 simultaneous HD streams, basic video conferencing, and general browsing. May struggle with 4K streaming or large file downloads while other household members are active.
• 100-300 Mbps: Meets the FCC's current broadband definition and handles most household needs comfortably. Supports 4-6 simultaneous HD/4K streams, multiple video conferences, and online gaming without noticeable congestion. This range represents the sweet spot of value for most American households in 2026.
• 300-1,000 Mbps: Provides substantial headroom for even demanding households. Large file downloads complete quickly, 4K streams start instantly, and even bandwidth-intensive activities like game downloads (50-100 GB) and cloud backups proceed at speed. Households with 10+ active devices benefit from this tier.
• 1,000+ Mbps (Gigabit): Exceeds current household needs for most families but provides maximum future-proofing. Particularly valuable for households with home servers, professional content creators, or those who regularly transfer very large files. The primary benefit at this tier is not raw speed for any single activity but the ability to support many simultaneous high-bandwidth activities without any degradation.

Upload Speed: The Often-Ignored Half

Most consumers focus exclusively on download speed, but upload speed is increasingly critical for modern internet usage. Check your upload test result against these benchmarks:

• Under 5 Mbps: Marginal for video conferencing. Single video calls may work but will have reduced quality. Multiple simultaneous video calls will fail. Cloud backup of photos and files will be extremely slow.
• 5-20 Mbps: Adequate for one video call at a time and background cloud syncing. Will struggle if multiple household members attempt video calls simultaneously.
• 20-100 Mbps: Comfortable for most upload-dependent activities including multiple simultaneous video calls, cloud backup, and social media content uploading. This is the minimum recommended range for remote work households.
• 100+ Mbps: Required for professional content creation (video uploading, live streaming), operation of multiple security cameras streaming to the cloud, and hosting server applications from home. Typically only available on symmetric fiber connections.

Latency (Ping): Why It Matters Beyond Gaming

Latency measures the time (in milliseconds) for a data packet to travel from your device to the test server and back. While often associated with gaming, latency affects every internet activity:

• Under 20ms: Excellent. Ideal for competitive gaming, real-time trading, and video conferencing. Typical of fiber connections to nearby servers.
• 20-50ms: Good. No perceptible delay for video calls or gaming. Typical of cable and well-positioned DSL connections.
• 50-100ms: Acceptable for most activities but noticeable in competitive gaming and real-time applications. Typical of DSL at longer distances or congested cable nodes.
• 100-200ms: Problematic for real-time activities. Video calls may have noticeable audio delay. Gaming is playable but with competitive disadvantage. Typical of satellite internet (non-Starlink) or very congested networks.
• 200ms+: Severely impacts usability of real-time applications. Video calls have conversation-disrupting delays. Gaming is impractical for competitive play. If your wired connection shows 200ms+ latency, there is likely a network issue that your ISP should investigate.

Troubleshooting Slow Speed Test Results

If your speed test results are significantly below your plan speed (more than 25% lower on a wired Ethernet connection), work through these troubleshooting steps before contacting your provider.

Hardware Checks

1. Restart your modem and router. Power off both devices, wait 30 seconds, then power on the modem first and wait for all indicator lights to stabilize (2-3 minutes). Then power on the router. This clears temporary memory issues and forces a fresh connection to your provider's network. This single step resolves the majority of speed issues.
2. Check your Ethernet cable. Damaged or low-quality Ethernet cables can limit throughput. Ensure you are using Cat5e or Cat6 cable (printed on the cable jacket). Replace any cable that shows visible damage, tight bends, or crushed sections. A $5 cable replacement has fixed many "slow internet" complaints.
3. Verify modem capabilities. If you own your modem, verify it supports your plan speed. A DOCSIS 3.0 modem may be limited to 300-600 Mbps even on a gigabit cable plan — you need DOCSIS 3.1 for full gigabit cable speeds. For fiber, ensure your ONT (Optical Network Terminal) supports your plan tier; older ONTs may limit speeds on newer, faster plans.
4. Test at the modem/ONT directly. Bypass your router entirely by connecting your computer's Ethernet cable directly to your modem or ONT. Run a speed test in this configuration. If speeds are now normal, your router is the bottleneck — it may need a firmware update, a factory reset, or replacement with a model that supports higher throughput.

Network Environment Checks

1. Scan for bandwidth-consuming devices. Log into your router's admin panel (typically 192.168.1.1 or 192.168.0.1) and check connected devices and their bandwidth usage. Security cameras, smart TVs with auto-play features, and devices downloading updates can consume significant bandwidth without your awareness.
2. Check for malware. Infected devices can consume bandwidth for botnet activities, cryptocurrency mining, or data exfiltration. Run a malware scan on all computers on your network if speed issues persist after hardware troubleshooting.
3. Verify DNS settings. Slow DNS resolution can make internet feel sluggish even when raw speed is adequate. Try switching to Google DNS (8.8.8.8, 8.8.4.4) or Cloudflare DNS (1.1.1.1, 1.0.0.1) in your router settings to see if responsiveness improves.

When to Contact Your Provider

Contact your ISP if: (1) Speed tests over Ethernet directly to the modem/ONT consistently show less than 70% of your plan speed. (2) You experience speed degradation that started suddenly and persists after modem restart. (3) Speed tests show high latency (100ms+) or significant jitter (latency variation of 20ms+) on a wired connection. (4) Speed drops dramatically during specific hours, suggesting node congestion (for cable customers). When calling, have your speed test results ready with dates, times, and the test tool used — this data helps technicians diagnose the issue efficiently rather than running through a generic troubleshooting script.

Speed Test FAQ

Which speed test is most accurate?

No single speed test is universally "most accurate" — they all measure slightly different things. For a balanced view, run both Ookla Speedtest (which tends toward optimistic results since it selects low-latency servers) and Fast.com (which tests against Netflix CDN servers, representing typical streaming performance). If the two results agree within 15%, you have a reliable measurement. For the most independent measurement, use Google's M-Lab test, though it may show lower numbers because M-Lab servers are deliberately not optimized for peak throughput.

How often should I run speed tests?

For ongoing monitoring, run a speed test once per week at roughly the same time. This creates a baseline that makes it easy to spot when performance degrades. If you suspect a problem, run tests multiple times per day for 2-3 days to establish a pattern. Many speed test apps (including Ookla's) can schedule automatic periodic tests, building a performance history without manual effort.

My Wi-Fi speed test shows much less than my plan speed — is my ISP cheating me?

Almost certainly not. Wi-Fi speed is limited by your router's capabilities, the distance and obstacles between your device and the router, interference from other Wi-Fi networks, and the Wi-Fi standard your device supports. A gigabit internet plan will not deliver gigabit speeds over Wi-Fi in most home environments. Test over Ethernet to confirm your actual internet speed. If the Ethernet test shows your full plan speed, the "slow" Wi-Fi result indicates a home network issue (router placement, Wi-Fi standard, interference) rather than an ISP problem. Consider a mesh Wi-Fi system or Wi-Fi 6E router to improve wireless coverage.

Do speed tests use my data cap?

Yes, speed tests transfer real data and count against any data cap on your plan. A typical speed test transfers 200-500 MB of data. Running multiple tests per day could use 1-3 GB, which is insignificant against most caps (1-1.25 TB) but worth noting if you are tracking data usage closely. If you are near your cap limit, avoid running speed tests until your billing cycle resets.

Can my ISP detect and boost speed during a speed test?

Technically, ISPs could detect traffic to known speed test servers and prioritize it, a practice sometimes called "speed boost" or "turbo." This was more common before net neutrality regulations. While specific practices vary by provider and jurisdiction, using multiple speed test tools (including non-obvious ones like Fast.com and M-Lab) helps detect such manipulation. If Ookla shows 500 Mbps but real-world downloads from Steam, Microsoft, or Apple consistently max out at 300 Mbps, your connection may be experiencing prioritization of test traffic.