Recover, Rest, or Retire? The Cold Email Domain Decision Tree

First: Stop Sending, Then Diagnose

The most expensive mistake in domain management is diagnosing mid-flight. Every additional send from a struggling domain digs the reputation hole deeper and muddies your test results. Pause cold volume on the suspect domain (leave warmup running), then work the five tests in order, together they take under an hour.

The Five Tests

• •1. Blacklists. Run the domain and its sending IP through the Blacklist Checker (20+ lists). A Spamhaus listing is a different disease than a clean sheet, it changes the whole tree. Minor lists (UCEPROTECT L3 and similar) matter far less than people fear.
• •2. Authentication. One pass through the Deliverability Checker, a silently broken SPF (the 10-lookup limit is the classic) or a DKIM key someone rotated wrongly imitates a burned domain perfectly, and fixes in an afternoon.
• •3. Seed placement. Send 5-10 plain, link-free test emails to fresh Gmail and Outlook seeds you control. Read the verdicts from headers with the Header Analyzer. Inbox at low volume = reputation is bruised, not dead. Spam even at trickle volume, on both providers = serious damage.
• •4. Bounce autopsy. Pull the last 2 weeks of bounces. Mostly 5xx user-unknown = a list problem wearing a domain costume, fix verification (see the catch-all policy). Mostly 4xx deferrals = throttling, not death.
• •5. Warmup read. Warmup networks report placement too. Warmup emails inboxing while cold emails junk = content/volume problem. Warmup itself junking = deep reputation damage.

Branch 1, Throttled: Keep the Domain, Change the Pace

Signature: clean blacklists, passing auth, seeds land (maybe delayed), bounces are 4xx deferrals clustered on one provider. The provider is rate-limiting you, not rejecting you. The fix is pace, not rehab: cut daily volume 50%, widen sending windows, check you're not over ~30 sends/inbox/day, and, if the deferrals cluster on Microsoft, read my Outlook piece, because unmatched Google→Microsoft sending throttles first and junks second. Resume normal volume over a week.

Branch 2, Damaged: The 30-Day Recovery Plan

Signature: no major blacklists, auth passes, but seeds junk at low volume and reply rates cratered. The domain's reputation took a real hit, usually a bad list, a volume spike, or spam complaints. Recovery that works:

• •Days 1-10, full rest. Zero cold sends. Warmup only. Fix the root cause now: re-verify the list, correct the volume discipline, repair auth. Rest without a root-cause fix is a scheduled relapse.
• •Days 11-17, trickle. ~5 sends/inbox/day, only to your cleanest, most-engaged segment (past repliers' lookalikes, freshly verified, provider-matched). Plain text, no links, no tracking.
• •Days 18-24, ramp. 10-15/day if bounces stay under 2% and a mid-week seed test shows inbox. Any regression → back one step, not zero.
• •Days 25-30, verify. 20-25/day, full seed test on both providers, blacklist re-check. Passing = the domain re-enters the rotation. Failing = you've reached the retirement branch honestly.

Branch 3, Retire: When Rehab Costs More Than Redeployment

• •Spamhaus (SBL/CSS) listing, or relisting after a successful delist, the underlying signal keeps refreshing.
• •Spam-trap evidence, pristine traps especially; that reputational stain doesn't launder.
• •A failed 30-day cycle, you gave it an honest recovery and placement still fails.
• •Economics, a sending domain costs ~$12 and two weeks of warmup (or arrives prewarmed). A second month of rehab while pipeline stalls costs far more. Retire without guilt.

The Real Lesson: Domains Rarely Die of Natural Causes

Almost every domain that reaches this decision tree was pushed, an unverified list, a volume spike, tracking pixels into Microsoft tenants, mixed catch-alls, or auth quietly broken for a month. Run the autopsy honestly and fix the practice, not just the domain, or your fresh domain inherits the same fate on schedule. The operators who stop losing domains aren't better at recovery, they're better at never needing it.