← back to blog

When Telegram flags your IP and not your account

telegram ip-reputation account-limits troubleshooting

When Telegram flags your IP and not your account

The first line in my triage notes says: ask them to turn wifi off. It sits above every question about the account.

That one instruction closes more tickets than everything else in the file put together. Somebody opens with a long paragraph about being banned, switches the handset to mobile data because I asked, and the account he was writing to me about loads before he has finished typing the message.

Nothing was wrong with the account. Something was wrong with the road it kept arriving on.

I run managed Telegram hosting on dedicated hardware in Singapore, real handsets on real carrier SIMs, so I get the pre diagnosed version of this problem several times a week. The diagnosis is usually wrong, and it is wrong in an expensive direction.

The two repairs point opposite ways

This distinction earns its keep for one reason. The fix for an address problem is to move. The fix for an account problem is to stop moving.

Get it backwards and you do not merely fail to fix anything. You take a network fault and convert it into a permanent mark on the account’s own history, because the flailing is itself a signal that gets recorded.

That conversion only runs one way. There is no equivalent mistake in the opposite direction, which is why the order you test in matters more than how much you know.

The test that comes before everything else

Take one affected account. Put it on a network that shares nothing with the one it is failing on. Mobile data on a handset, with every proxy and VPN switched off, is the cleanest version of this. A hotspot from a phone on a different carrier is second best.

If it works immediately, the account is healthy and you can stop examining it.

If it fails on the second network in exactly the same way, the problem is riding along with the account, and no infrastructure purchase is going to shake it loose.

Thirty seconds, and you can undo all of it. Almost nobody runs it first.

Signals that point at the address

Simultaneity is the strongest one. Every account on that connection hits the same wall inside the same short window, despite being different ages and belonging to different people who use them for completely unrelated things. Accounts do not coordinate. A four month old account and one registered in March have no reason to fail together on a Tuesday unless something they share changed, and the only thing they share is the exit.

A brand new account failing at registration is the second. If a signup dies during verification, or in its first hour, that account has no behaviour available to be judged on. Where it came from is very nearly the only fact in play, so the verdict lands on the address by elimination.

The third is the mobile data result you already have. Same number, same device, same person, one different road, immediate success. One variable moved and the outcome flipped.

There is a fourth that is softer but still useful. Address problems tend to bite at connection or at registration. They do not usually wait until you try to send one specific message to one specific stranger.

Signals that point at the account

It follows you. New network, new device, fresh install, a friend’s wifi in another building, and the restriction is still sitting there. That is history the account is carrying around, not something it picks up locally.

It is isolated. One account limited while four others on the same connection carry on as normal. Addresses do not aim at individuals.

It attaches to an action. Logging in works fine. Messaging people who never contacted you is what stops, or adding members, or forwarding at volume, while nothing about the connection has changed at all.

It runs on a clock. If you are told to wait a specific number of seconds and the wait clears on schedule, that limit is behavioural by construction. A countdown does not care which address you serve it from.

Why an address carries other people’s behaviour

A shared address is one surface with a lot of people standing on it, and Telegram sees the surface.

Cheap proxy pools, most consumer VPNs, and any ordinary office router put your traffic and a stranger’s traffic on the same exit. From the outside those are one stream. A single neighbour running something aggressive earns the address a reputation, and everyone behind it inherits the result, including the account that has only ever read two channels.

The part people miss is that an address is not blank when you receive it. Blocks get handed back, reassigned and rotated constantly. The address you were given this morning belonged to somebody else last month, and whatever they did on it is now your starting position. It is invisible to you and there is nobody to ask about it.

So a proxy that behaved perfectly for a year can turn inside a fortnight with nothing changing at your end. The address stayed the same. The tenants moved.

A new account on a bad address is a different situation entirely

Here is the asymmetry that produces two people getting opposite results on the identical address.

An old account arrives holding evidence. Months of ordinary use, a stable device, contacts who reply, nobody reporting it. Weigh that against one mediocre address and the address is a minor input.

A new account arrives with none of that. The address becomes the entire case, because it is very nearly the only fact that exists about the account.

So one address can be perfectly fine for a three year old account and completely unusable for a registration attempted an hour later on the same connection. Both of those results are correct.

I have been caught by this from the wrong side. A customer kept telling me his signups were dying, and I kept telling him the connection was clean, because my own account had been sitting on it happily for months. My account told me about my account. It said nothing whatsoever about a number registering for the first time that afternoon. He burned through four of them before I accepted the difference.

The rule that falls out: the worse your address, the older the account you need to survive it. Registering is the most address sensitive thing anybody does on this platform.

The common failure looks like diligence.

You hit an address wall, so you try a different proxy. It half works. Then a VPN, then a friend’s hotspot, then a provider trial, then mobile data. By the evening that account has appeared on six networks in three countries.

Telegram watches where an account shows up over time. An account that lives on one connection for months is boring, and boring is the whole goal. An account surfacing on a new network every forty minutes has the exact shape of a stolen session being passed around, and that shape is worth reacting to.

So the churn becomes evidence in its own right. The history now reads erratic, and it does not reset when you eventually land somewhere clean. You started with an address problem and you now own both kinds.

There is a quieter loss on top. You have destroyed your own diagnosis. After eight addresses in one day you cannot say which of them was fine, and neither can anybody else.

Change the address once, on purpose, then leave it alone long enough to learn something from it. That means days.

What resolves the address half

Stop sharing the surface. Everything else here is detail.

An address that only your account exits through has a reputation assembled from your own behaviour, which makes it something you can manage. A shared one is a lottery you re enter every morning without ever being shown the odds.

Carrier mobile addresses sit above the rest for a structural reason. The same ranges carry very large numbers of ordinary people on their phones, so treating those ranges as disposable costs Telegram real users.

Stability still beats category. An unremarkable address you keep for six months outperforms a premium address you swap every fortnight, consistently.

Then wait. An account that has just moved is an account that has just changed networks, which is a small flag by itself. Give it a week of dull, consistent use before you decide whether the move worked.

What resolves the account half

Considerably less, and I would rather say so than pretend.

If the limit is on a timer, serve it and stop doing whatever started it. Infrastructure has no effect on a countdown.

If it is a genuine restriction, the appeal goes through Telegram, and you are asking a system to reverse its own decision, which is a different activity from negotiating with a person.

If it is a young account restricted for volume, the honest answer is that it needed a slower first month, and there is no way to add one after the fact.

The part I cannot give you

Telegram does not publish which of the two it is, and it never will, because that would mean publishing the detection logic. Everything above is inference from watching the same patterns repeat across a lot of accounts. It is reliable enough to act on. It is still inference.

There is also a real grey zone. A marginal address underneath a two week old account, where the address alone would be survivable and the account alone would be unremarkable, and correcting either one by itself changes nothing you can observe. Those cases exist, they eat a week, and anybody who claims they can always tell the two apart is selling you something.

telegramvault.org is managed Telegram hosting on dedicated Singapore hardware, real handsets on real carrier SIMs, with each account sitting on an address that nothing else exits through. Onboarding starts with a conversation about your setup, and the code TGYT gets you a discount when you start.

Get new guides and videos first — join the Telegram channel.

need infra for this today?