What does warming a new sending address look like from the customer side?
From the customer side, warming is two published lines and a daily authentication row. Our own notification mail has not moved onto that view.
It looks like two published lines and a daily authentication row. It does not look like a promise about the inbox. Our own notification mail has not moved onto this view, so what follows is the screen we insist on reading before we point customer mail at a new address, not a diary of a warmup we have finished.
WarmySender sells warming and deliverability. We spend our days on the operator side of a new address: volume that starts small, engagement that has to be real, a reputation that is easy to spend and slow to earn back. Being the customer of a sending API is a different seat. The question we care about from that seat is narrower. What number will stop our mail, is that number written down before we hit it, and can we see authentication failures per day rather than at the end of the month.
The lines that are already public
The deliverability page publishes the same figures the enforcement uses. We checked it on 23 September 2026. It was last updated on 22 September 2026.
Bounce rate warns at 2% and pauses at 4%. Complaint rate warns at 0.05% and pauses at 0.08%. No automatic step runs until the rolling window holds at least 100 delivered-or-bounced sends, so a handful of failures on a quiet domain cannot trip it. One step per evaluation. The panel's red line is the rate at which enforcement begins, so a domain sitting on the line is not a warning you can sit on.
Standing does not clear the moment a rate dips back under the line. It clears at the next review. A rate that oscillates across a line would otherwise flap the standing, and a badge that changes every few minutes is a badge nobody reads. We want that delay. A warmup that "recovers" because one good hour pulled a tiny sample under the line is not a recovery.
Bounces are not one bucket. A hard bounce, a soft bounce, a block, and a suppression are four problems with four fixes. A bounce rate with no classes under it tells you something is wrong and nothing about what to change. Hard bounces and complaints add the recipient to the suppression list. Sending to a suppressed address is refused rather than delivered.
The page also states the volume warning in plain numbers: a domain with no history that sends fifty thousand messages on its first day is indistinguishable from a compromised one. That sentence is the customer-side version of what we already tell people who warm with us. Raise volume gradually. The fifty thousand is their line for "this looks like a compromise," not a target and not a plan of ours.
The daily authentication row
Receivers mail a daily report of the messages that claimed each domain: which addresses sent them, and whether they aligned. The view we will read is one row per domain per day. It carries how many messages claimed the domain, how many aligned, how many failed, and the failure rate for that day. Beside the days, the addresses that passed as the domain and are not ours are listed on their own.
That split is the part a warmup hides if you only watch your own sends. Mail that fails alignment is usually our own configuration: something sending as the domain without the right signature. Mail that passes as the domain from an address we do not send from is somebody else, or a service we forgot. Both show up in a daily row. Neither shows up in a bounce rate.
We are not going to publish a policy that tells receivers to quarantine failures until those rows have been quiet. A policy change is the deliverability edit you cannot undo quickly, because receivers keep the old record until their cache expires. The daily row is how we decide we are not early.
What we will not pretend
AgentiSend is built by the same team as WarmySender. Read this as the view our own team will use, not as an independent deliverability study. The product is live and sending real mail. That is not a placement rate for our domains. We have no week of our own notification mail on this view, because that mail has not moved. A warmup company publishing a plan and calling it a result would be the failure we warn customers about. This note is the plan.
The other honest limit is what the lines do not measure. A domain under 2% bounces and under 0.05% complaints can still land in a spam folder. The published lines are when a human should look and when sending pauses. They are not an inbox-placement figure, and we will not invent one to fill the gap. Simulated sends are excluded from the rates, which is correct for testing and useless as evidence. Evidence is delivered and bounced mail from a real address, in a window of at least 100, read against numbers we could recite before we sent.
Two habits sit next to the lines. Publish the authentication records and keep them published. Handle bounces the day they happen, not at the end of the month. Neither is new to us. The new part is that the customer-side screen and the enforcement read the same numbers, so we do not find out from a support ticket that we crossed 4%.
The figures and the two tables are on the deliverability guide. The product is at agentisend.com.