Cold Email Deliverability Dropped? How to Diagnose and Recover
“Replies died and nobody can tell me why” is the message that starts half my client engagements. Deliverability drops have a diagnosable cause — the trick is checking in the right order instead of guessing. This is the exact sequence I run.
Placement test
Seed-test by provider. Is it Gmail-only, Microsoft-only, or everywhere? The pattern narrows the cause immediately.
Auth check
Verify SPF/DKIM/DMARC on every affected domain — a DNS change or expired key breaks fleets silently.
Blacklist scan
Check the lists that matter (Spamhaus first). Ignore vanity-list noise.
Volume & list audit
Pull the last 30 days: volume spikes? New unverified list? Bounce or complaint jump? There's your spark.
Reading the placement pattern
- Everywhere at once → authentication broke or a serious blacklisting. Check DNS first — it's the five-minute fix that looks like a disaster.
- One provider only (say, Gmail fine, Outlook junking) → provider-specific reputation. Cut volume to that provider, keep engagement traffic, consider routing those recipients through a matching ESP pool.
- Some domains, not others → domain-level burn. Isolate them, check what those domains did differently (list, volume, copy), begin recovery or rotation.
The recovery playbook
Rotation: the honest exit
Rehabilitating a hard-burned domain is sunk-cost theater. The professional move is a rotation pipeline: spare domains always warming in the background (the +25% headroom in fleet math), so a burned domain swaps out same-day with zero campaign downtime. Prevention beats all of it — most drops trace back to one of the nine causes of spam placement, and all nine are avoidable.
The five-minute checks that find most drops
Before the placement test finishes, run the DNS. In my experience roughly a third of "deliverability collapsed overnight" cases are a record somebody changed: a DKIM selector rotated by the provider and the old CNAME left in place, an SPF record that picked up an eleventh lookup when a new tool was added, a DMARC record moved to p=reject on a domain where a third of the mail was never aligned. The domain checker shows all of it on one screen for a domain; the DMARC checker and SPF checker go one level deeper on the record that looks wrong.
Then read a real bounce. Not the summary in your sending tool — the actual SMTP response. A 550 5.7.26 from Gmail is authentication and nothing else will fix it. A 5.7.1 is policy, which on Gmail usually means reputation. A 5.7.606 from Microsoft is a banned IP and is theirs to fix, not yours. The bounce code decoder maps any code to the responsible layer, which tells you which of the steps above to skip.
Reading the numbers your sending tool already has
Three ratios, read over the last fourteen days per domain, separate a real drop from a copy problem. Bounce rate above 2–3% means the list is the cause, not the infrastructure, and no amount of warmup fixes a list. Reply rate falling while open rate holds means placement is fine and the message stopped landing — a copy or targeting problem, not a deliverability one. Reply rate and open rate falling together, on one provider, is the placement drop this article is about.
Ignore open rate as a number on its own; privacy proxies inflate it. Use it only as a direction, and only alongside replies.
The mistakes that turn a dip into a rebuild
Sending harder to "push through" a dip. The receiver is already sampling your mail at a lower trust level; more of it accelerates the verdict. Switching all inboxes to a new copy at once, which turns a placement question into a placement-plus-copy question you cannot untangle. Rotating domains at the first sign of trouble, which throws away warmed assets and starts the whole fleet's reputation from zero at the moment the receiver has just learned to recognise the pattern. And killing warmup because "the domain is warm now" — a drop is exactly when the warmup layer's positive engagement is buying you room.
Do the opposite of each: halve the volume on the affected domain, hold the copy, keep the domain, move warmup back to the early 60% share, and test placement again in four days. Most dips that were caught by the numbers above recover in one to two weeks that way. The ones that do not are the ones where a cause on the list above was never fixed.
Set up so the next drop is caught in a day
The real cost of a drop is the time between it happening and you noticing. A weekly placement test per domain and a daily automated check of DNS, blacklists and MX health takes that from weeks to a day. The domain monitor runs the automated part; the reputation risk calculator turns your bounce, complaint and volume numbers into a risk score before the receiver does, which is the cheapest warning you will get.
Everything above is diagnosis. To confirm a fix worked you need placement data — seed accounts across the major providers, plus the DMARC aggregate reports read as a trend rather than a wall of XML. GlockApps is what I run for both.
See GlockApps →Affiliate link — it costs you nothing and help keep these tools free. Everything I use →Tools for this
Domain checkerEvery delivery record for a whole list.DMARC checkerSee whether your policy enforces anything.SPF checkerValidate a record and see who it authorises.Bounce code decoderEvery SMTP and status code, decoded.Frequently asked questions
How long does email reputation recovery take?
Mild damage typically recovers in two to three weeks of reduced, high-engagement sending. Moderate damage takes four to six. Hard burns — sustained spam placement plus major blacklistings — often never fully recover, which is when rotation beats rehabilitation.
Should I stop sending completely when deliverability drops?
Cut cold volume 70–80%, but don't go fully silent — a total stop followed by a restart looks like a new spike. Keep warmup-level engagement traffic running so providers see continuous, positive behavior while you fix root causes.
When should I give up on a domain and rotate it out?
If placement hasn't meaningfully improved after three to four weeks of proper recovery, or the domain hit Spamhaus, retire it. A warmed replacement reaches full capacity in under a month; a hard-burned domain can stay damaged for a year.
Is it my copy or my deliverability?
Look at opens and replies together, per domain. If opens hold and replies fall, the mail is landing and the message is the problem. If both fall on one provider while the others hold, it is placement on that provider. If both fall everywhere on the same day, it is DNS or a serious blacklisting. A placement test settles it in an hour; guessing from the campaign average does not.
Can I fix deliverability by changing the subject line or removing links?
Only if the problem was content, which is the least common cause on a properly built fleet. Stripping links and spam-trigger words from a message that is failing on reputation changes nothing except making the message worse. Diagnose first. If the placement test shows Gmail junking mail that passes authentication and is not listed anywhere, reputation is the cause and volume plus warmup is the fix.
How do I know recovery is actually working?
Placement test every four or five days on the same seed set. Recovery shows as inbox placement climbing on the provider that dropped, usually over one to three weeks at reduced volume. If two consecutive tests show no movement, something on the checklist is still broken — go back to DNS and blacklists rather than waiting longer.