Down today, or gone for good?
A service being unreachable for an hour and a service closing permanently are completely unrelated events. They get searched for in almost the same words, and confusing them wastes a lot of worry.
A service being unreachable for an hour and a service closing permanently are completely unrelated events. They get searched for in almost the same words, and confusing them wastes a lot of worry.
Search volume for "is Roblox shutting down" is consistently high, and has been for years, during which Roblox has not come remotely close to shutting down. The same is true of Minecraft, of Discord, and of almost every large platform with a young audience.
These searches look like outage searches and are not. Someone typing "is X down" wants to know whether to wait ten minutes. Someone typing "is X shutting down" has heard that a service they use is closing forever, and wants to know whether to panic. This site answers the first question. This page is about the second, because the two get conflated constantly and the answer is almost always reassuring.
| An outage | A shutdown | |
|---|---|---|
| How long it lasts | Minutes to hours | Permanent |
| Warning | None | Usually months |
| Who announces it | Often nobody | Always the company, in writing |
| How it starts | A change, a dependency, a load spike | A commercial decision |
| What you should do | Wait | Export your data |
The crucial asymmetry: A company that is closing a service tells you. It has to. There are subscriptions to wind down, data-protection obligations to meet, refunds to process and app stores to notify. A shutdown that arrives without an announcement is vanishingly rare, and when it does happen the company has usually collapsed rather than decided.
Three triggers account for almost all of them.
This is the one that overlaps with the subject of this site. When a platform is unreachable for several hours, a plausible-sounding explanation fills the gap, and "they are shutting down" is the most dramatic available. Roblox's three-day outage in 2021 produced exactly this, at enormous volume, for a company that was publicly listed and growing.
The tell is timing. Shutdown talk that begins during an outage and nowhere else is a story invented to explain the outage.
Both produce genuine uncertainty and both get over-read. A company cutting staff is usually trying to stay solvent, which is the opposite of closing. An acquisition sometimes does lead to a product being retired - But through an announced migration, months later, with notice.
Some shutdown rumours are simply evergreen fabrications, recycled every year or two with a new date attached. They spread fastest among younger audiences and on platforms where a screenshot travels further than a correction. The date passing changes nothing; the rumour returns with a new one.
The single best filter: A genuine shutdown announcement always contains a date you can put in a calendar. If the claim has no date, or the date keeps moving, it is not an announcement - It is a rumour with a number in it.
Google Reader, Google+, Vine, Wunderlist, Skype, Windows Phone. The pattern is remarkably consistent:
Nothing in that sequence resembles a site being unreachable this afternoon.
Worth mentioning because it shows up in the same search patterns: Some "did X shut down" queries are about physical facilities rather than services. "Did CERN shut down" is asked about the Large Hadron Collider, which genuinely does shut down - For scheduled maintenance periods lasting months, announced years in advance, and entirely unrelated to whether any website is reachable.
We do not cover those, and a status page would be the wrong answer if we did. A particle accelerator going into a planned maintenance stop is not an outage, and nothing an uptime monitor measures would tell you anything useful about it.
Export first and read the terms second. Export tools get slower and queues get longer as the date approaches, and some have quietly stopped working before the advertised deadline.
Check what happens to anything you paid for - Remaining subscription time is usually refunded or pro-rated, and purchased content sometimes is not, which is worth knowing before rather than after. And take the read-only period seriously as a deadline rather than as extra time; it is common for exports to be unavailable during it.
Then, genuinely: Check it is not just down. Every year a substantial number of people conclude a service has closed when it was unavailable for an afternoon, and the cost of checking is about thirty seconds.
No. It is a publicly listed company that has never announced a shutdown, and the question spikes whenever the platform has an outage or a hoax circulates. A real shutdown would be announced by the company in writing, with a date and a data-export path, months in advance.
Look for the company's own written announcement with a specific date and an export tool. Every genuine shutdown has all three. A rumour typically has a dramatic claim and either no date or a date that keeps moving.
Because a long unexplained outage leaves a gap that people fill with the most dramatic available explanation. Shutdown talk that starts during an outage and appears nowhere else is a story invented to explain the outage.
Usually not, if you act. Genuine shutdowns come with an export tool, partly because data-protection law in many jurisdictions requires one. Export early - Those tools get slow and queues get long as the deadline approaches, and some stop working during the final read-only period.
Typical outage durations by cause, why most incidents resolve in minutes, and the warning signs that an outage is going to be a long one.
The real causes of website outages - Configuration pushes, BGP withdrawals, expired certificates and dependency cascades.
A practical order of tests for working out whether a website is genuinely down or the problem is on your side - And what each result rules out.
Why most real outages are partial, how regional and feature-level failures hide from monitoring, and how to tell what is actually broken.