A typical fleet runs like this: an account works until it is lost. Convenient and expensive — because it is always lost at the wrong moment.
Three states instead of two
Not «working» and «dead» but three: working, resting, retired. The middle one is what is usually missing.
Resting means a few days of no actions after a restriction or a burst of load. The account stays connected and does nothing. Same mechanism as warm-up, run backwards.
When to rest one
After a first restriction. After a run of send failures. After a week spent at the top of the normal range. And simply by rotation — if the fleet allows, let every account rest regularly.
When to retire one
After a second or third restriction in a row. An account collecting them regularly is already flagged, and from there it only spoils your statistics. On the difference between «restricted» and «finished», see what to do about a restriction.
Why you need spare capacity
So that taking an account out does not stop the task. A few accounts above the required number are not waste — they are what makes rotation possible at all. How many for a given volume: how many accounts you need.
What it buys
A steady fleet instead of a sawtooth. Attrition becomes predictable and enters the budget as a planned line — see how long an account lives and why cheap costs more.
Buy accountsaged, with full session files