Which proxy type is keeping your accounts past thirty days?

I have same problem for six month and i want to compare with other people before i spend more money.

Accounts die between day 20 and day 35, almost all of them, very consistent. Not day 2, not day 90. Something happen around one month. I use residential rotating, one session per account, sticky for 30 minute.

My theory is the rotating is the problem — the account see a new city every session and after some week it add up. But mobile proxy is 8x the price and i not want to pay this for 60 account if the problem is something else.

So tell me what you run and how long your account live. Residential, isp, mobile, datacenter if you are brave. And say the number, not "they last fine".

Thirty days is the trial period. You're not being banned, you're being graduated.

Numbers, since you asked for them. About 180 accounts over the last year, all the same platform, all warmed the same way, so the only variable is the connection.

Proxy	n	Alive at 30d	Alive at 90d
Residential rotating, sticky 30m	61	54%	19%
Residential static (same IP always)	44	86%	71%
Mobile, shared, 5 accounts per IP	39	90%	77%
ISP / datacenter-residential hybrid	36	72%	48%

So your theory is right, but the fix isn't mobile — it's static. The difference between rotating and static residential is 32 points at 30 days and it costs almost nothing to change. Mobile buys you another four points over static residential at several times the price, and in my data that difference isn't clean enough to be worth what it costs unless the accounts are individually valuable.

Caveat: my static residential is genuinely static, not "sticky for 24 hours". That distinction is where a lot of providers get vague and it's the whole variable.

This is exactly what i need, thank you. 54% at 30 day is almost same as me.

How you get truly static residential, all provider i ask say sticky.

You buy it as a product — it's usually sold as ISP proxies or static residential, priced per IP per month rather than per GB. If they're quoting you bandwidth, it's rotating with a timer on it.

The proxy is not what kills the account and the table above is measuring something else.

What kills it is inconsistency between the things the platform can see, and the IP is only one of maybe fifteen. A rotating proxy produces inconsistency reliably, so it correlates with death, and everyone concludes the IP mattered. Move to static and the inconsistency moves elsewhere — a timezone that doesn't match the IP, a locale string that doesn't match either, a fingerprint that's identical across sixty accounts on sixty different IPs, and a login pattern with the regularity of a cron job because it is one.

And 30 to 35 days is not a coincidence. That's a review window, not an event. Something about the account was decided in week one and acted on in week five.

I don't disagree with any of that and my table still holds — I held the fingerprint, the timezone and the warm-up constant across all four rows. Whatever else is going on, changing only the proxy moved survival by 32 points.

Both things are true. The IP isn't the only signal, and it's a signal you can fix on a Tuesday for £40.

Fair. My objection is to the conclusion people draw from tables like yours, not the table.

Actually one thing about mobile proxy for who is thinking about it — the shared one is not bad like people say.

I have 5 account on one mobile ip and is fine, more than one year some of them. In real life a mobile tower have thousand of people, so five account on one ip is nothing, is normal. The mistake is 50 account on one ip, this is not a tower anymore, this is a farm.

My cost is 22 dollar month for one mobile proxy, divide by 5 account is 4.4 each. Is not 8x if you share correctly my friend.

Ok this change the maths completely. I was calculating one proxy one account.

how do you decide which 5 go together

Never account that follow each other or post same thing. Five stranger, is all.

FWIW if you go static, monitor whether the IP is still yours. Providers reassign, and the failure is silent — you keep connecting, the account keeps working, and then one day the account is on an IP that has been three other people this month.

Log the egress IP on every session and alert when it changes. Two lines of script and it'll explain a death you'd otherwise write off as random.

Silent reassignment. Of course.

Six week later, small update.

Moved 30 account to static residential, keep 30 on rotating as control because is the only way to know. At day 40: static 26 alive, rotating 17 alive. Is smaller sample than Marta but same direction.

Also i log the egress ip like PipelineRat say, and already catch one — provider give me new ip on day 11 without telling. That account die on day 29. So maybe my static number should be better even.

Not moving to mobile yet. If the 5-per-ip thing hold i try it for the accounts that matter.