How To Recover When Deliverability Suddenly Drops

Most cold email operators only think about deliverability when something breaks. By the time a mailbox score drops below 100 percent, or a domain starts bouncing, the damage is already compounding. This is the playbook I run every time a client pings me with that exact sentence: "my emails just stopped landing."

The recovery work starts before the next campaign goes out: verify the domain, clean the list, and route around high-friction recipient infrastructure. If your drop is tied to enterprise gateways, pair this playbook with the Mimecast filtering cleanup workflow so you are not sending a recovered mailbox back into the same problem.

The 48-hour rule

If a mailbox score drops below 100 percent, or any platform pulls it out of an active campaign, the answer is the same every time. Stop sending on every mailbox for 3 to 5 days. Not "reduce volume." Not "pause the worst one." Stop. All of them. Turn off sending, turn off warm-up, and pull every mailbox out of every sequence.

This sounds aggressive. It is. The reason it works: when one mailbox starts failing, the other mailboxes on the same sending infrastructure are usually next. ESPs and inbox providers look at the account, not just the address. If you keep sending from the healthy ones while one burns, you create a paper trail that says "this sender has a problem." Pausing everything resets the conversation.

Why you always need 50 percent headroom in mailbox count

This is the rule every new agency skips. Whenever you buy mailboxes, buy at least 50 percent more than you need today. If you need 10 mailboxes for your current daily volume, buy 15. If you need 20, buy 30.

People push back on this constantly. "That's wasteful." It is not. It is the difference between recovering in 3 days and recovering in 3 weeks. When one mailbox goes bad, you swap in a backup, your daily volume stays the same, and the affected mailbox gets to rest without your whole operation grinding to a halt.

The math is simple. The hard part is buying the extra mailboxes before you need them, because nobody wants to spend money on infrastructure they are not using yet. That is the trap. The backup is not optional. It is the seatbelt.

The health score thresholds

Most cold email platforms give you a health score for each mailbox. The common thresholds:

• Above 95 percent: green light. Send normally.
• 85 to 95 percent: amber. Reduce daily volume by half and run a warm-up pass.
• Below 85 percent: red. Pull out of campaign, run pure warm-up only, no sends.
• Below 70 percent: dead. Do not try to revive. Replace the mailbox.

The mistake I see is teams that watch a mailbox slide from 95 to 80 over a week and do nothing. By the time they react, the inbox provider has already flagged the underlying sending reputation, and the score will keep dropping even if you pause it. Reactive management is too late.

Set up two alerts per mailbox: one at 95 percent (warning), one at 90 percent (action required). When the second alert fires, the mailbox comes out of campaigns that day.

The recovery loop for one mailbox

When a single mailbox goes red, the recovery loop is:

1. Pause all sends from this mailbox. Pull it from every sequence. This includes any automated follow-ups.
2. Pause warm-up. Counterintuitive, but warm-up that keeps the mailbox "active" while the score is dropping is just more activity on a flagged sender. Stop the noise.
3. Wait 3 to 5 days. No sends, no warm-up, no logins from new IPs. Let the inbox provider's signal decay.
4. Restart warm-up only. Day 4 or 5, turn warm-up back on. Light volume: 5 to 10 sends per day to known-good addresses.
5. Watch the score for 4 to 7 days. If it climbs back above 95 percent, the mailbox is recoverable. If it stalls at 85 to 90 percent, the reputation is poisoned; do not put it back in production.
6. Reintroduce to campaigns slowly. If the score recovers, start the mailbox at 30 percent of normal daily volume. Ramp to 60 percent on day 2, 100 percent on day 3.

The full cycle is 7 to 12 days. That is the cost of one bad mailbox. Compare this to the cost of replacing it: a new mailbox takes 14 to 21 days to warm up from scratch. So if recovery is even possible, recovery is faster than replacement.

Mailbox health score drops
        │
   Score ≥ 95% ? ── Yes ─→ Keep sending, watch daily
        │ No
        ▼
Pause ALL sends now · pull from sequences
        ▼
Pause warm-up (yes, warm-up too)
        ▼
Wait 3 to 5 days — zero activity
        ▼
Restart warm-up only (5–10 sends/day)
        ▼
   Score back to 95%+ ?
        │
   No ──┤── Yes
        │        │
        ▼        ▼
Mailbox is      Reintroduce slowly:
poisoned.       Day 1: 30% volume
Replace it.     Day 2: 60% volume
Start new       Day 3: 100% volume
warm-up.

Why this is so hard to do under pressure

The reason most teams fail at this playbook is not technical. It is psychological. When deliverability drops, the pressure is to "do something." Reply to more leads. Send more emails. Volume it up. That is the exact opposite of what works.

Every extra email you send from a flagged mailbox tells the inbox provider the same story: this sender is not stopping, even though their reputation is broken. The response is to escalate, not to forgive.

The real fix is to stop, wait, and recover. That requires you to absorb a week of lower volume. Which is exactly why you bought the backup mailboxes. If you did not, this is when you find out, and the cost is 2 to 4 weeks of recovery across your whole sending operation.

What to check on your list before you restart

Before you put a recovered mailbox back in campaigns, run these three checks on the list it will send to:

1. Bounce rate. If your last campaign had more than 3 percent hard bounces, the list is part of the problem. Validate it. Remove any role addresses (info@, contact@, support@) and any catch-all domains. Hard bounces are the single fastest way to destroy a sender reputation.
2. Engagement mix. Aim for at least 40 percent of the list to have opened or replied to one of your emails in the last 90 days. Cold lists with zero engagement history are fine for new mailboxes; they are poison for a recovering one.
3. Spam trap risk. Run the list against a verification service that flags known spam traps. Free lists and scraped lists almost always contain some. Removing them is non-negotiable.

If your list fails any of these, fix the list before you restart. A clean list on a damaged mailbox will still damage the mailbox. A dirty list on a clean mailbox will damage the clean mailbox.

Domain and authentication checks

If multiple mailboxes are dropping at once, the problem is usually not the mailboxes. It is the sending domain. Run these checks:

• SPF record. Must include every service that sends on behalf of your domain. Most ESPs publish the exact include: line you need. If you added a new sending tool and forgot to update SPF, that tool's emails will look forged.
• DKIM signing. Every sending service should have its own DKIM record published in your DNS. Test it by sending yourself an email and inspecting the headers. If dkim=pass is missing, you have a DKIM problem.
• DMARC policy. Start with p=none so you can monitor. Move to p=quarantine once you have clean reports. Never jump straight to p=reject without 30 days of clean data.
• Custom tracking domain. If you are using the ESP's default tracking domain, every email looks like every other ESP customer's email. Set up a custom tracking subdomain (mail.yourdomain.com). It costs nothing and it isolates your reputation.

A sending domain with broken authentication is a sender that inbox providers cannot trust, regardless of how clean the list is or how good the copy is. Fix this once and you stop having "random" drops.

The restart protocol

Day 1 after recovery: send at 30 percent of normal volume to your cleanest, most engaged segment. Watch reply rates, watch bounce rates, watch the health score. If everything is green, ramp.

Day 2: 60 percent volume. Same segment plus a slightly broader engaged list.

Day 3: 100 percent volume. Resume normal sending patterns.

If at any point the health score dips below 90 percent, restart the recovery loop. Do not push through. The whole point of the playbook is to fail fast and recover fast, not to recover by sending more.

The daily health dashboard you actually need

You do not need a 40-tab dashboard. You need a single sheet with one row per mailbox and these columns:

Mailbox              Health  Bounce%  Reply%  Status  Action
rep1@yourdomain.com    96%     0.8%    3.2%   Green   None
rep2@yourdomain.com    88%     4.1%    1.1%   Amber   Halve volume
rep3@yourdomain.com    79%     7.2%    0.4%   Red     Pause + recover
rep4@yourdomain.com    99%     0.3%    4.1%   Green   None

Update it daily. If a row hits amber, action it the same day. If a row hits red, action it within the hour. This is the entire job.

What to do this week

1. Audit your current mailbox count. If you are running above 70 percent utilization, buy backups before you need them.
2. Set up the daily dashboard. Even if it is just a Google Sheet. The data already exists in your ESP; you just need to surface it.
3. Verify your SPF, DKIM, and DMARC records. Use mxtoolbox.com or mail-tester.com. Fix anything broken.
4. Test one full recovery loop on a low-traffic mailbox. You do not want your first time running this playbook to be during a real crisis. Practice it when the cost of a mistake is small.
5. Document the playbook. One page. Save it where the whole team can find it. The day deliverability drops is not the day to invent the process.

Deliverability recovery is not glamorous work. It is the same loop every time: stop, wait, clean, restart, watch. The teams that survive a drop are the ones that have run the loop before, on purpose, when it did not matter. Everyone else is learning it during a fire.

Questions operators actually ask

How do I fix email deliverability when it suddenly drops?

Pause all campaign sends first, then diagnose authentication, mailbox health, bounce rate, and domain reputation before restarting. The fastest path is not more volume; it is less activity, cleaner lists, and a controlled restart to your most engaged segment.

How long does it actually take to recover a poisoned mailbox?

Seven to twelve days for the full loop, assuming the underlying sender reputation is recoverable. Three to five days paused with zero activity, then four to seven days of warm-up before it is safe to re-enter production at 30 percent volume. If the score stalls at 85 to 90 percent after a week of clean warm-up, the mailbox is not coming back. Replace it and start a fresh 14 to 21 day warm-up on the new one.

Should I keep sending from my healthy mailboxes while one is in recovery?

No. Pause everything tied to the same sending infrastructure. ESPs and inbox providers look at the account, not the address. Sending from the survivors while one is burning builds a paper trail across all of them. This is the move that turns a one-mailbox problem into a five-mailbox problem in 72 hours.

Is the daily health dashboard really worth the setup cost?

Yes, and the setup cost is small. One Google Sheet with one row per mailbox and six columns. Twenty minutes to build. The reason teams skip it is the same reason they skip backups: it is invisible until the day it matters.

What if my bounce rate is below 3 percent but replies are dead?

The list is not the problem; the copy or the targeting is. Bounce rate below 3 percent means inbox providers are accepting the sends. Reply rate near zero means the prospects are not opening them, or they are opening and ignoring. Both point away from infrastructure and toward message-market fit. Pause only the campaigns that are dead; do not pause the mailboxes.

How do I check email deliverability and domain reputation?

Check Google Postmaster Tools for Gmail reputation, Microsoft SNDS for Outlook signals, your ESP mailbox health score, and a seed test or header check for SPF, DKIM, DMARC, and custom tracking-domain alignment. One dashboard should show health score, bounce rate, reply rate, and current send status for every mailbox.

When do I replace a mailbox versus trying to recover it?

Below 70 percent health, or stalled at 85 to 90 percent after a full clean warm-up cycle. At that point the sender reputation is poisoned at the domain or IP level, and recovery takes longer than warming a fresh mailbox. Do not chase a reputation you cannot read. Replace.

Does turning warm-up on or off help recovery?

During the pause phase, turn it off. Warm-up keeps the mailbox generating activity on a flagged sender, which deepens the hole. Restart warm-up only at day 4 or 5, after the quiet period, and only at 5 to 10 sends per day to known-good addresses.

What is the single most underrated inbox signal to watch?

Microsoft SNDS data. Most operators check Google Postmaster and stop there. SNDS gives you the Outlook.com delivery picture two to four days before Gmail's score catches up, which is the only early-warning window you have to recover without losing the domain.

How can I improve email domain reputation after a drop?

Remove risky leads, fix every authentication failure, restart at low volume, and send first to recently engaged contacts. Do not add new cold segments until the health score, bounce rate, and reply rate have stayed stable for several days.