How We Test VPN Speed

In order to give you the most accurate, up-to-date results possible, we at VPN.com have developed an advanced system which accounts for several different variables in network infrastructure like no other site to date.

See, while many other websites in the VPN review space claim to have accurate speed tests, all of their results only measure how long it took a packet to travel from the VPN server in a specific location to wherever the writer of that article happened to run the speed test.

Not only that, but if their results were posted using the old Flash-based Speedtest.net tool provided by Ookla, you can’t even be sure they’re accurate due to Flash’s inability to properly decode encrypted TCP traffic.

This is why we set out to change the model of how speed test results are collected and calculated. First, it all starts with the new beta HTML5-based speed testing tool over at Speedtest.net. Because HTML5 isn’t crippled by the same limitations as Flash, all the results gathered via this new updated method are proven to be the most accurate taken on any review site to date.

The Fiber Optic Difference

While other review sites might only test their VPNs using a 100Mbps line (some even go as low as 25Mbps), neither one of these speed tiers are able to represent the true capacity of a VPN speed operating at its peak.

In order to get the most accurate results possible we’ve tested every VPN featured on our domain using a fiber optic, 1GB/s simultaneous upload and download line.

This means that when we say you’re getting the fastest VPN on the market with fast VPN speed, you know you’re truly getting the fastest of the fast – no questions asked.

Local Server Locations

To increase accuracy even further we then continued to refine our process down to the granular level, guaranteeing the results you see will reflect what you should expect from a provider when connecting from within that country, province, or state. See instead of posting results that reflect how long a packet takes to travel from a VPN server, to the Speedtest.net node, and then back to our local testing location in Portland, Oregon (as all other review sites have), we found the fastest server located in each city and test from there. VPN speed vary and depends on location and time too.

Once the global test cities were decided, it was time to begin the process of selecting which Speedtest.net server we would use to test the VPNs. For example when testing potential server contenders in Sao Paulo, Brazil, we ran a test using a local VPN node in that city to get results for every server that Speedtest.net provides.

After those results were in we took the top three servers from each city and gave them three more runs in a “final round”, after which the server node with the fastest average result would be selected as the winner and the server we would test off for that city moving forward.

Off-Peak Hours

To remain as unbiased and fair as possible, we wanted to do whatever we could to give every VPN provider tested in this project the best chance of success when it came to putting their fastest foot forward.

This meant testing their servers only during what is known as “off-peak” hours, or around 6 AM local.

Statistically 6 AM is the time when most VPN providers see the lightest amount of load on their servers and VPN speed enhances, which means that anyone who is trying to see what a network is capable of at its best should only be running tests during that hour.

Luckily that’s just what we’ve done, making sure to only post results that were recorded a 6AM local for every location featured in the document.

Three Test VPN Speed Average

Once the VPN speed is connected to the fastest server in a specific region, we then run six tests in total to determine the overall average. The first three are dry runs which connect to the closest server here in Portland, designed to gather data for a control of what the unrestricted non-VPN speed looks like at the hour we tested locally.

Next we run the VPN test itself, which again consists of three back-to-back tests from the same server. After the results are in we average them together and that brings us to the final result!