How We Test VPN Speed: Our Full Methodology
Why Most VPN Speed Tests Are Unreliable
The VPN review industry has a credibility problem. Most websites publish speed test results that are impossible to reproduce, use inconsistent methodologies, or worse, fabricate numbers entirely. At VPNTex, we believe that transparency in testing methodology is just as important as the results themselves.
This article documents our exact testing process so that anyone can replicate our results. We also explain common mistakes in VPN speed testing and why many published benchmarks are misleading.
Our Testing Infrastructure
All VPN speed tests at VPNTex are conducted from a dedicated testing environment in Oslo, Norway. Our test machine runs on a direct 1 Gbps fiber connection from a major Norwegian ISP, with a consistent baseline speed of approximately 940 Mbps download and 480 Mbps upload.
The test machine runs a clean installation of Windows 11 Pro with no background applications, no browser extensions, and no other network-active software. We use a wired Ethernet connection exclusively — WiFi introduces too many variables to produce reliable measurements.
We test each VPN using its native desktop application with default settings. This means we test whatever protocol the VPN selects automatically, which is typically WireGuard or a WireGuard-based protocol for modern VPN clients.
Testing Procedure
Each VPN is tested across five server locations representing different distance categories. These locations are Stockholm (short distance), Frankfurt (medium distance), London (medium distance), New York (long distance), and Tokyo (very long distance). We believe this range captures performance across realistic use cases.
For each server location, we perform the following measurements. Download speed is measured using Ookla's Speedtest CLI tool, taking the average of three consecutive tests with a 30-second pause between each test. Upload speed is measured using the same procedure. Latency is measured as the round-trip ping time to the VPN server. DNS resolution time is measured using the dig command against the VPN provider's DNS servers.
We run the complete test suite three times per day: morning (08:00 CET), afternoon (14:00 CET), and evening (20:00 CET). This accounts for peak-hour congestion, which can significantly affect VPN speeds. Our published results are the average of all nine measurements (three tests × three time slots) for each server location.
What We Measure and Why
Download Speed
Download speed is the most important metric for most users. It determines how quickly web pages load, how smoothly videos stream, and how fast files download. We express download speed in Mbps and also calculate the percentage of baseline speed retained through the VPN.
Upload Speed
Upload speed matters for video calls, cloud backup, file sharing, and content creation. Many VPN reviews neglect upload speed entirely, but for users who regularly upload data, this metric is critical.
Latency
Latency (ping time) affects the responsiveness of your internet connection. Low latency is essential for video calls, online gaming, and general web browsing snappiness. We measure latency in milliseconds and report the average across our test runs.
Common Testing Mistakes
Mistake 1: Testing on WiFi
WiFi speeds fluctuate based on interference, distance from the router, and network congestion. A VPN speed test conducted over WiFi cannot isolate the VPN's impact from WiFi variability. Always test on a wired connection.
Mistake 2: Single Test Runs
A single speed test is essentially meaningless. Network conditions vary from second to second. Running only one test and publishing the result is like measuring the temperature once and declaring it the daily average.
Mistake 3: Ignoring Time of Day
Internet speeds vary dramatically throughout the day. Testing a VPN at 3 AM and comparing it to one tested at 8 PM will produce wildly different results that have nothing to do with the VPN's actual performance.
Mistake 4: Using Different Baseline Connections
Comparing VPN speeds tested on different internet connections is invalid. A VPN tested on a 100 Mbps connection cannot be meaningfully compared to one tested on a 1 Gbps connection. All tests must use the same baseline.
| Testing Factor | Our Approach | Common Mistake |
|---|---|---|
| Connection type | Wired Ethernet only | WiFi testing |
| Tests per server | 9 (3 tests × 3 time slots) | Single test |
| Time coverage | Morning, afternoon, evening | One time of day |
| Baseline connection | Fixed 1 Gbps fiber | Varying connections |
| Protocol | Default (VPN's choice) | Cherry-picked protocol |
| Server selection | Fixed 5 locations | Random or best-case |
How We Calculate Scores
Our overall speed score for each VPN is calculated using a weighted average across all server locations. Short-distance servers receive higher weight (40%) because they represent the most common use case. Medium-distance servers receive 35% weight, and long-distance servers receive 25% weight.
The final speed score is expressed as a percentage of baseline speed retained, ranging from 0 to 100. A score of 85, for example, means the VPN retains 85% of your original connection speed on average across all test locations and times.
Reproducibility
We publish our raw test data for every VPN review. You can download the spreadsheet containing individual test results, timestamps, and server locations. We encourage readers and VPN providers to attempt to reproduce our results — if our methodology is sound, similar results should be achievable under similar conditions.
Updating Our Results
VPN performance changes over time as providers add servers, optimize routing, and update software. We retest our top-rated VPNs quarterly and update results accordingly. If a VPN's performance changes significantly between scheduled reviews, we note this prominently on the relevant review page.
Our commitment is to honest, reproducible, and transparent speed testing. If you have questions about our methodology or want to suggest improvements, please contact us.
