Which Facts About a VPN Provider Are Still True Next Year

Everything you can learn about a VPN provider has a shelf life, and the useful move is to sort facts by that shelf life before you weigh any of them. Ownership, place of establishment, funding model and protocol support change on the scale of years. Location counts, application behaviour, promotional terms and whether the service currently reaches any given site change on the scale of weeks. A search with a year in it is really a question about the second group, which is exactly the group that cannot be answered durably.

This page publishes no ranking and no winner. What it offers instead is the sorting rule, so you can tell which sentences in anybody’s write-up are worth remembering.

The long-lived facts

These are the ones worth researching properly, because the work does not have to be repeated every few months.

Who owns the operator, and what else that owner runs. Corporate ownership does change, but it changes as a discrete, documented, findable event rather than continuously. A parent company’s other lines of business tell you what the group is for, and that is stable information.

Where the company is established, and therefore who can compel it. This is a legal fact about a registered entity, not a marketing claim, and it moves only when the company reorganises. It is also the fact most often replaced in marketing by an emotional substitute — a flag, or a phrase about a privacy-friendly country. What the category of question actually looks like is set out in VPN jurisdiction and the Five Eyes, explained.

How the service makes money. Subscriptions, a paid parent product, advertising, data, or something unstated. A funding model is the most predictive durable fact available, because it tells you which way the incentives point when a decision is difficult.

Which protocols and standards it supports. Protocol support turns over slowly and is publicly documented. It is also one of the few technical facts with real consequences for you, since the protocol in use affects reconnection behaviour, overhead, and how the tunnel copes with a network that moves under it.

Its addressing practices. Whether exits are shared or reserved, whether the second address family is carried or discarded, whether name resolution is handled inside the tunnel. These are architectural decisions and they persist across redesigns.

Its documented history. Incidents, legal demands that became public, changes of ownership, and how the operator described each at the time. History does not expire; it only accumulates.

Whether an independent examination exists, and what it covered. The existence and scope of a published examination is a permanent fact. Its reassurance value is not, which is a distinction worth keeping — see what a VPN audit is.

The short-lived facts

Every item here can be true when written and false a few weeks later, with nothing visible on the page to warn you.

Counts of locations and machines. Inventory numbers move constantly and were never comparable between providers anyway, since nobody publishes what a counted location consists of.

Whether the service currently reaches any particular site. This is the live state of an ongoing back-and-forth between two parties who both keep changing their approach. Treat any statement about it as a report on one moment.

Application behaviour, defaults and platform quirks. Clients get rewritten. Most of your day-to-day experience lives in the application, and the application is the least stable component in the whole arrangement.

Commercial terms. Prices, bundles, trial lengths, refund windows and what a given tier includes are administrative details that get revised without announcement.

Observed performance. A throughput result belongs to a path at a time, not to a brand. Why it cannot be inherited from an article is the subject of the fastest-VPN claim and how to test it yourself.

Support quality. It depends on staffing, and staffing changes.

The rule this produces

Any statement in the second group needs a date attached to the sentence, not to the page, and the page’s date is usually the only one you get. A year in a title is an editing timestamp: it tells you somebody touched the file, not that anybody re-established the perishable claims underneath.

There is a practical version of this. When you read a write-up, mark each claim long or short. If what convinced you sits in the short group, you have not learned something about the provider — you have learned something about a moment that has already passed. If what convinced you sits in the long group, the write-up’s age barely matters, and you can verify most of it yourself from the operator’s own documents.

What a year-stamped page could honestly contain

Not nothing. It could carry the long-lived facts with sources, plus the short-lived ones clearly labelled as observed on a stated date, plus an explicit note about which of them the author actually checked rather than restated. That page would be useful and would age visibly rather than silently, which is the point.

What it cannot honestly contain is an ordering, because an ordering is a single number computed from both groups at once. Blend a fact with a five-year life and a fact with a five-week life and the result inherits the shorter one, while looking as settled as the longer.

Why this site does not publish the list

We have not used these services, so we have no observations of our own to date-stamp. A ranking assembled from other people’s pages would be a mix of long-lived facts we did not verify and short-lived ones we could not, presented as a conclusion — and a reader has no way to tell which parts came from where. That is not a cautious omission; publishing it would actively mislead, because the format signals testing that never happened.

The honest contribution from a site like this is the sorting rule, which is durable in the way the rankings are not.

Using it

Research the long list once, from the operator’s own filings, policies and documentation, and keep your notes. Re-check the short list at the moment you need it, on your own connection, and never carry a short-list claim over from last year’s reading. Then work through the rest of the decision in the order set out in how to choose a VPN.