Ask Your Cold Email Vendor How They Decide When to Rotate a Mailbox, a Domain, or the Whole Provider

When a cold email campaign's numbers crash, the fix depends entirely on which layer broke: one mailbox, the whole sending domain, or the provider and IP pool behind it. A vendor who can't tell you their diagnostic order for telling those apart is guessing, and guessing costs you weeks of wasted volume.

Domain reputation and provider reputation are not the same thing

Mailgun's deliverability documentation draws a hard line between the two. Domain reputation ties to your DKIM-signed sending history and persists no matter who you route mail through. Provider or IP reputation is separate, resets on every switch, and needs a fresh 2 to 4 week warmup each time. SendGrid's guidance draws the same split, with formal warmup schedules built around it.

Switch providers to fix a burned domain, and you have just re-warmed on top of the same bad reputation, on a new clock. Nothing got solved.

How Maily actually runs this

MDR's email-ops agent, Maily, checks in a fixed order when a campaign craters: mailbox health first (warmup score, bounces, disconnects), because it is cheapest to rule out. Then DNS and domain, SPF, DKIM, DMARC alignment, plus a blacklist check against Spamhaus DBL and SURBL. Only if both come back clean does a per-provider placement test run. Copy gets looked at last, because infrastructure problems dress up as bad copy constantly, and rewriting copy that is actually landing in spam wastes everyone's time.

The thresholds are fixed, not vibes. Bounce rate under 2% is fine, 2 to 5% is a warning, over 5% means pause volume and audit the list. Complaint rate under 0.1% is fine, 0.1 to 0.3% is a warning, over 0.3% and Gmail starts filtering, an immediate pause. A single spam trap hit gets investigated on sight, no threshold needed.

A blacklisted domain gets paused and a delisting request goes out with it, and it does not return to rotation on a guess, only once it is actually delisted. That is also why roughly half of MDR's mailbox capacity sits in warmup at any given time, plus a backup pool, so a degrading mailbox's volume can shift to healthy ones while it gets investigated, instead of running to failure on a fixed schedule.

This is a different axis from the list decay covered in our other vetting piece, where UNICARGO's 12 mailboxes across 6 domains came up. That post is about contacts going stale. This one is about the sending infrastructure itself.

Here is the honest gap. There is no single published rotation policy document on a shelf. What exists is the diagnostic order and thresholds above, run by hand on every campaign, real practice rather than a brochure, and worth saying plainly instead of pretending a formal SOP exists.

What to ask before you sign with anyone

• When reply rates drop, what do you check first, and what's the actual order after that?
• What are your real bounce and complaint rate numbers for pausing volume, not a general sense of "too high"?
• If a domain gets blacklisted, what happens next, and how do you decide it's safe to send from again?
• How much spare mailbox capacity do you keep so a degrading sender can be pulled out while it's investigated, instead of run to failure?

Ask me those same four questions. I'll give you the same answer I just gave you here.