FIELD NOTES
A personal record of travel, networks and small failures
TRAVEL NOTE

My ISP Could See the VPN—What Mattered Was Whether It Could Block It

The sexual-health discussion I wanted to read had disappeared behind an age-verification screen. I was an adult, but uploading identification to open a forum thread felt disproportionate. A private window made no difference, so I connected a familiar VPN, moved the connection outside the UK and reloaded the page. The discussion appeared immediately. Ten minutes later, while checking my broadband dashboard for an unrelated outage, I noticed that my laptop had spent the entire session communicating with one unfamiliar server. The websites were no longer visible, but the tunnel itself clearly was.

That left me with the question I should have asked before installing anything:

Could my internet provider tell that I was using a VPN?

Yes. But seeing the VPN was not the same as seeing what I did through it.

The short answer

At that point, the question was no longer whether the mobile provider could read the health discussion. It could not, because the VPN connection never became usable.

The Part My ISP Could Still See

This question has become more urgent in the UK since stronger age-assurance requirements took effect in July 2025. VPN use rose sharply afterward, while later government research found that privacy and access to restricted services were common reasons people—particularly younger users—turned to them.

The policy debate soon shifted from age checks to circumvention. Could networks identify VPN users? Could VPN traffic be restricted? And would using one create a conspicuous record of its own?

The same confusion appears repeatedly in public discussions. People ask whether an ISP can still “track” them through a VPN, when they are usually combining three separate concerns: whether the ISP can see the destination, recognise the tunnel or read what happens inside it.

Once I separated those questions, the answer became much clearer.

Without a VPN, my broadband provider carries traffic toward each website or online service. HTTPS protects most page contents, but the connection can still reveal useful information about where the traffic is going.

Turning on a VPN changes that route. My laptop creates one encrypted connection to a VPN server, and the ISP carries that connection instead.

The ISP can see the server’s IP address, when the connection begins, how long it lasts and roughly how much data passes through it. What it no longer receives is a straightforward list of the websites carried inside the tunnel.

From the provider’s side, opening the health discussion, checking email and reading the news all became encrypted traffic moving toward the same server. It could see that I was using a tunnel. It could not open that tunnel and read the forum post, messages or pages inside it.

That distinction solved my original privacy concern at home.

Then I changed networks.

When Recognising the VPN Became Enough

I left the house with the established VPN still connected. As the laptop moved from home Wi-Fi to my phone’s mobile hotspot, the tunnel dropped.

I pressed reconnect.

The app stayed on “connecting,” failed and suggested another server. I tried the Netherlands, then Germany, then France. Each attempt ended the same way.

The provider’s website still opened over the ordinary mobile connection, so the service itself had not simply disappeared. The app displayed hundreds of available locations. None of that helped the tunnel complete.

At that point, the question was no longer whether the mobile provider could read the health discussion. It could not, because the VPN connection never became usable.

The important fact was that a network does not need to decrypt a VPN tunnel to interfere with it. Recognising the server address or the shape of a familiar VPN protocol can be enough to slow, interrupt or reject the connection.

Large providers often use well-known endpoint ranges. Their protocols also produce identifiable connection patterns. Research has shown that ISP-level systems can classify many conventional OpenVPN and WireGuard connections by examining structural signals rather than reading the encrypted contents.

That explained the contradiction in front of me. The major provider offered an impressive server list, but every route I selected announced itself in a way the network could reject.

The ISP still could not see the private page.

It did not need to. Preventing the tunnel from forming achieved the same practical result.


The Connection That Reached the Page

That was when I opened OnlydogVPN.

Instead of working through another map of countries and protocol menus, I selected a preset for restrictive networks and connected.

The status changed almost immediately.

I reopened the browser. The health discussion loaded. So did the information page linked beneath it and the PDF I wanted to save before a medical appointment. Email continued syncing in another tab.

The result came first: the same mobile connection that had rejected several attempts from the established provider was now carrying a working encrypted route.

Only then did the design difference matter. The smaller app used HTTP/3-based transport with additional traffic obfuscation. Rather than presenting the same recognisable connection pattern, it was designed to resemble ordinary encrypted web traffic more closely.

I could not inspect the mobile provider’s internal classification rules, so I could not see the precise label applied to either connection. What I could observe was decisive: the conventional tunnel repeatedly failed, while the obfuscated connection opened the page and remained usable.

That changed the comparison.

Both services encrypted traffic. The first one never got far enough for that encryption to help me. The second established a route before the network could stop the task.

The preset also removed the guesswork that had wasted the previous ten minutes. I did not need to decide whether another country, port or protocol might be less recognisable. I described the situation through the preset, connected and returned to the page.

The browser—not the settings screen—became the test.

Why the Network Switch Mattered

Once the PDF had downloaded, I put the phone in my pocket and boarded the train.

The signal weakened between stations, returned and moved between mobile bands. The page paused briefly, then continued. The tunnel recovered without sending me back to a country list or reconnect button.

That was a smaller benefit than getting through the initial interference, but it completed the experience. Mobile connections rarely stay still. A VPN that works only until the next network change can turn a private browsing session into repeated troubleshooting.

The established provider had already required several manual attempts before the train arrived. With the smaller app, I stopped thinking about the route once the page opened.

That difference made the VPN worth keeping installed. It was not merely hiding a destination while conditions were ideal. It was preserving access while the underlying connection changed.

What My ISP Knew—and What It Did Not

The mobile provider still knew that my device was exchanging encrypted traffic with an external server. It could measure the timing and volume of that traffic.

It did not receive the sequence of websites, forum posts, emails and documents moving through the working tunnel.

That is the practical answer to the original question: an ISP can often recognise that a VPN connection exists without seeing what the user is doing inside it.

The distinction becomes important when the network begins interfering with VPN traffic. At home, the familiar provider worked because the tunnel was allowed to form. On mobile data, recognising that tunnel was enough to stop it.

OnlydogVPN solved the more immediate problem. Its obfuscated transport established the connection, its preset removed unnecessary trial and error, and its recovery kept the route alive as the network changed.

The service has fewer locations and a shorter public history than the largest VPN brands. Those differences matter when the goal is maximum geographic choice. They mattered far less when several countries on a larger server list all produced the same failed connection.

I had started by asking whether my ISP could see the VPN.

By the time the train reached my stop, I understood that this was only half the question. The ISP could see encrypted traffic, but it could not see the private page inside the working tunnel.

For the information I needed that afternoon, hiding the destination mattered—but establishing a tunnel the network did not immediately reject mattered first.

Questions this experience may leave you with

What was actually causing the problem?

At that point, the question was no longer whether the mobile provider could read the health discussion. It could not, because the VPN connection never became usable.

Why did the obvious fixes fail?

The same confusion appears repeatedly in public discussions. People ask whether an ISP can still “track” them through a VPN, when they are usually combining three separate concerns: whether the ISP can see the destination, recognise the tunnel or read what happens inside it.

What should you check first?

Both services encrypted traffic. The first one never got far enough for that encryption to help me. The second established a route before the network could stop the task.

What finally changed the result?

I could not inspect the mobile provider’s internal classification rules, so I could not see the precise label applied to either connection. What I could observe was decisive: the conventional tunnel repeatedly failed, while the obfuscated connection opened the page and remained usable.

What is worth remembering?

By the time the train reached my stop, I understood that this was only half the question. The ISP could see encrypted traffic, but it could not see the private page inside the working tunnel.