Account warmup

A brand-new account that posts, follows fifty people and sends messages within its first hour has no history at all. Real accounts do not behave that way: someone signs up, looks around, closes the app, comes back the next evening, follows a few things they actually care about. Warmup reproduces that arc deliberately, and it exists because platforms apply their tightest thresholds to exactly the window where an account has nothing to be judged on except its rate of action.

What a platform can see about a new account

Very little, which is the problem. In the first days the only signals available are the age of the account, the count and cadence of its actions, whether the actions reciprocate (do the people it follows follow back, do its posts get any engagement), whether a phone or email is verified, and the reputation of the device and IP it arrived on. An account with no history on a device with no history behind an address with a history of other accounts is being scored almost entirely on that last term.

A schedule that holds up

• Days one to three — open the app, scroll the feed, watch a few items to the end, open two or three profiles. No follows, no posts, no profile edits.
• Days four to seven — follow a small number of genuinely relevant accounts, like occasionally, and finish the profile: avatar first, bio a day later, link later still.
• Second week — first posts, still at low volume, published at times that vary by hours rather than minutes.
• Week three onward — the activity the account exists for, ramping gradually rather than stepping up to full volume in one day.

Session shape matters more than daily totals

Three sessions of four to twelve minutes spread across a day read very differently from one forty-minute block, even when the action counts are identical. Inside a session, the details that vary in real use are dwell time per item (roughly one to eight seconds, with occasional much longer stops), scroll velocity, and the fact that people re-read, scroll back up, and abandon. A session that consumes exactly twelve items at exactly the same pace every day is a schedule, not a habit.

The device needs a history too

A phone whose only installed app is the one being warmed, with an empty gallery, no saved Wi-Fi networks, no keyboard dictionary, a battery permanently at 100% and a screen that has never rotated, is not a used phone. It is a fixture. Warmup is worth doing on a device that has accumulated some ordinary state: a handful of unrelated apps, real uptime between reboots, at least one BSSID in its network history, a battery level that moves, and a locale, timezone and language that agree with the proxy region.

How to tell warmup is failing

Suppression is usually visible before a ban is. The observable symptoms, roughly in order of severity:

• Action blocks — "Try again later" on a follow or a like. You crossed a rate threshold; the account is now flagged rather than merely limited.
• Silent reversion — follows that appear to succeed and are gone an hour later.
• Reach collapse — posts showing near-zero accounts reached while the account looks otherwise healthy.
• Verification loops — a challenge on every login, then a demand for both email and SMS, then a demand for the same phone number a second time.
• Content not appearing under its own hashtags when viewed from a signed-out session on an unrelated connection.

The response to any of these is the same and it is counter-intuitive: stop entirely for 48 to 72 hours, then resume at half the volume that triggered it. Continuing at the same rate to "push through" a block converts a temporary limit into a permanent one, because the retry pattern itself is now the signal.

Errors precise enough to name

• Uniformity. Twenty accounts warming on the same schedule, with the same delays and the same follow count per day, have described a pattern more clearly than no warmup would have. Jitter of a few percent is not jitter — vary intervals and daily totals by tens of percent, per device.
• Warming on a device or IP already linked to other accounts. Perfect pacing on a shared fingerprint buys nothing at all.
• Completing the profile in the first ninety seconds: avatar, bio, link and pinned post before the account has ever scrolled a feed.
• Buying an aged account and logging into it on day one from a new device, a new country and a new IP. The age you paid for is discarded by the login that contradicts every session before it.
• A password reset or a continent-scale IP jump during warmup, both of which reset the trust the account had accumulated.
• Following back your own other accounts to make profiles look populated, which builds exactly the graph described in multi-accounting.

Automation is worth having for this only if it varies its intervals per device rather than replaying one timeline across a fleet — see RPA for what that requires in practice.