The Price of a VPN
At Daxdi, we've long believed that thinking of security as a zero-sum game isn't helpful.
It fosters a sense of helplessness, making people believe there's nothing they can do to protect themselves from all the nasty digital things in this world.
We see this mindset as especially dated when it comes to virtual private networks, or VPNs.
These simple, powerful tools keep your data protected within an encrypted tunnel.
Bad guys snooping on your traffic, ISPs, and spies can't see what you're up to when your VPN is active.
That said, using a VPN will have an impact on your internet connection, because doing so adds more steps to the process of transferring your data.
These extra steps generally degrade your internet speeds, simply by adding more fiber, more complexity, and more physical distance to the equation.
In exchange, however, using a VPN helps protect your data and personal security.
While speed is important, other factors such as value and privacy protections are critical.
As such, we tend to de-emphasize speed in our reviews, unless it is impressively good or abysmally bad.
But readers are, understandably, concerned about the impact a VPN will have on their web browsing experience.
And that's why we test.
Editors' Note: IPVanish, Encrypt.me, and Ookla are owned by j2 Global, the parent company of Daxdi's publisher, Ziff Davis.
What's the Fastest VPN, and Why Did it Change?
Each year, we run a massive slate of speed tests.
This year, we tested 39 VPNs in just a few weeks.
This data determines the fastest VPN overall for the year.
The Best VPN Deals This Week*
- NordVPN — $89 for Two-Year Plan With Bonus Gift (List Price $286.80)
- IPVanish VPN — $31.49 for One-Year Plan (List Price $89.99)
- Norton Secure VPN — $39.99 for One-Year Plan (List Price $79.99)
- TunnelBear VPN — $120 for Three-Year Plan (List Price $359.64)
- Surfshark VPN — $59.76 for Two-Year Plan (List Price $310.80)
- ExpressVPN — $99.95 for One-Year Plan + Three-Months Free (List Price $194.20)
- CyberGhost VPN — $87.75 for Three-Year Plan + Three-Months Free (List Price $506.61)
- Private Internet Access VPN — $69.95 for Two-Year Plan + Two-Months Free (List Price $310.70)
*Deals are selected by our partner, TechBargains
In the most recent round of tests, the results were mixed.
Hotspot Shield VPN led in two categories, chalking up the best download speed and latency test results.
Its upload speed results left something to be desired, however.
Surfshark was in second place in the download and latency tests, and first place in the upload test.
Our testing favors the download scores above the rest, so Hotspot Shield VPN is crowned the fastest VPN of 2020, but Surfshark isn't far behind.
That's notable in a year that saw overall scores improve across the board, suggesting that this may not be a settled contest for long.
Of course, we at Daxdi do not rest on our laurels.
We continue to update those reviews and test entirely new products throughout the year.
In the past, we have seen upsets of reigning champions.
Perhaps that will happen again this year, so watch this space.
The top results of our tests are in the chart below.
These are the top performers of the 39 we tested, so there's a chance your VPN of choice isn't on the list.
How We Gather Our Data
You can read a lot more about how we test VPNs and, importantly, how our testing has changed over the years in our feature appropriately titled How We Test VPNs.
The gist of it is that we have a much faster connection for testing since we started many years ago, and have adjusted our methodology in a way we feel better represents all the data we gather.
When we review VPNs, we use a custom tool provided by Ookla.
Built off the online Speedtest tool, which you can use yourself, it measures three important metrics: latency, download speeds, and upload speeds.
Those are the three areas we look at when comparing VPN speed performance.
Latency is a measurement of time between when your computer sends a request and when it receives a response.
It's also called ping time.
Lots of things can affect latency; the distance your request physically travels through fiber has a big impact, for example.
Latency is measured in milliseconds, however, so even a large increase may not be noticeable to the average user.
Latency is important when playing video games over a VPN, as lower latency means a more responsive experience with less lag.
Download and upload speeds measure how much data is moved over your internet connection in a certain amount of time.
These are both measured in Megabits per second (Mbps).
This is the metric ISPs and wireless companies tout when they claim their connection is "fastest."
When we test VPNs, we try to get a sense for the impact a service has on internet performance by finding the percentage change between using the VPN and not using the VPN for several speed measurements.
We run ten tests without the VPN active and find the median of the results.
This is our baseline.
We then do the same thing, but with the VPN active.
We then find a percent change between the speed test results with and without the VPN.
The bigger the change from the baseline performance, the more impact the VPN has on your internet speeds.
Previously, we dropped the highest and lowest results from our tests and found the mean of what remained.
We've since moved to taking a median measurement instead, as this preserved some of the up-and-down results we sometimes observe.
It also removed an element of human error and allowed us to double the number of tests we run on each VPN service.
We have tested each of these services in as repeatable a manner as possible, but there are limitations inherent in our testing.
The biggest issue is that our testing is carried out from the Daxdi labs in New York City.
If we performed the same tests on a similar network in a different part of the country, we'd almost certainly get different results.
Networks are also fickle things and even small changes can greatly affect the results of a speed test, so while we have controlled for as many variables as we can, it's hardly a "sterile" network environment.
We think of these tests as more of a snapshot of performance that establishes a replicable metric for measuring each service.
The goal is for comparison between services, not individual evaluations of speed.
Your mileage with these services will almost certainly vary somewhat from ours.
How We Interpret Our Data
Because our testing generates three data points for each VPN service, we must interpret that data in a way that makes sense to us.
We cannot simply take the top ten results for each category and call it a day (although we do break down those results below).
Nor can we simply pick one metric and use it alone to determine the fastest VPN.
When sorting through VPNs, we favor download Speedtest results above all others.
We assume that most people reading are major consumers of online content.
Reading the news, streaming movies, using BitTorrent, or listening to music on the web all require that your device pull down data more or less continuously.
We then look at upload results and then latency results, in that order.
We also apply a sniff test to our results.
While we try to hew as close to pure, objective numbers as possible, if a VPN has two great results and one absolutely terrible result, we cannot include it on the list.
To arrive at our list of the fastest VPNs, we first compiled a list of all the VPN services that had results better than the median in any of our three categories: download speed, upload speed, and latency.
For each category that the service beats the median, it gets one point.
This resulted in eight across-the-board winners, listed in order of their download results: Hotspot Shield VPN, Surfshark VPN, HMA VPN, TigerVPN, Mullvad VPN, NordVPN, Encrypt.me, and IPVanish VPN.
These are the services that made the cut for our list.
Another nine products beat the median in two two categories.
They are, in order of upload score: PureVPN, IVPN, TorGuardVPN, ShieldApps Cyber Privacy Suite, ProtonVPN, F-Secure Freedome VPN, Avast Secure Line VPN, AVG Secure VPN, KeepSolid VPN Unlimited.
IVPN stands out on the list for its remarkably low upload results, and its near miss on the latency results.
ProtonVPN was also a near-miss in one category, and it has strong scores across the board as well.
Consider these two as unofficial choices for fastest VPN.
Perhaps you disagree with the framework we use to interpret our test results.
With that in mind, we break down the top ten results for each of the three categories below.
What VPNs Have the Best Latency?
Once tallied up, we found that our latency results have a median of 69.6 percent.
Of all the VPN services we tested, these have latency scores less than or equal to the median.
They are shown here in descending order—Hotspot Shield takes top marks here.
Previous testing showed some services with a 0.0 percent increase in latency.
No more.
In fact, this year's testing had far more variation in latency than the previous year.
Despite that, the median result was significantly lower than last years' median of 87.5 percent.
This was a trend across all the results.
What VPNs Have the Best Download Speeds?
Our results show a median percent change in download speed of 68.5 percent.
The VPNs with results better than or equal to the median are shown here in descending order.
Hotspot Shield wins again here.
In this test, we measure how much using a VPN decreases download speeds over a gigabit connection.
We're confident that, given the speeds available to our connection, the greatest limiting factor in the results is the VPN service.
But because our baseline connection is so fast, even the best service we tested had a sizeable impact.
We assume that most readers won't be using a connection like ours, and we would likely see a smaller percent change in their results.
However, we believe our tests found the fastest possible speeds for a VPN connection in our geographic area.
Here's a hypothetical scenario: In one round of testing, we had a download baseline of about 926Mbps which was reduced to about 160Mbps when the VPN is running.
In our model, that works out to be about an 80 percent reduction in speed.
It is possible that with a connection half as fast as ours—say, 500Mbp—that the VPN could reduce speeds to the same figure of 160Mbps.
This would yield a smaller percent change, but still have the same bandwidth for the VPN connection.
This is all to illustrate that our test results are intended to be used for comparison between each other, not as an absolute rating of speed.
It's fair to say that a VPN that has a lower score in our tests performed better than other VPNs in the same test.
It's not fair to say that a service we found to reduce our download speeds by 99 percent will reduce your download speeds by 99 percent.
What VPNs Have the Best Upload Speeds?
Our speed tests found a median percentage change of 63.6 percent for upload speeds.
The VPNs that had results as good or better than the median are shown here in descending order.
Surfshark leads in this category.
Again, we're confident in the outcome of these tests and believe it to be useful for comparison, but it is not intended to reflect what every consumer will experience.
And again, this years' results were an overall improvement over the previous.
Cheating and Data Compression
Because we report how we test VPNs and because we do so using one of the most popular speed test tools available, there's some obvious concern that a disreputable company might attempt to game the results.
A nefarious VPN company could, perhaps, detect when an Ookla test is running and return bogus results.
We spoke with the engineers at Ookla when designing our tests, and they share our concerns about gaming their test.
They've told us that the company works to prevent this kind of cheating, taking active measures to fool would-be cheaters.
We defer to the experts at Ookla to prevent companies from gaming its test.
Because we make our methodologies public, it is also possible that VPN companies could tweak their networks to obtain better results.
For example, a VPN company could endeavor to rent server space as close to our offices as possible, hoping that the shorter distances will yield better speed test results.
There's little we could do to prevent a company from doing this.
That said, speed testing is only a small part of our reviews, and even an outstanding speed test result wouldn't do much to raise the score of an otherwise mediocre product.
Beyond Windows
When we evaluate a VPN's speed performance for this roundup, we use the speeds recorded with the VPN installed on a Windows machine.
We do this because most of our readers use Windows.
Bear in mind, however, that performance on one platform does not equate performance on another.
For information on how a specific VPN...
