Cold Email Deliverability Checklist for Automated Campaigns

A cold email deliverability checklist for automated campaigns needs six checks: authentication, mailbox health, lead quality, send controls, reply exits, and failure alerts. For a 12-person B2B agency protecting a 10.6% reply rate, I would pause the campaign if authentication fails, bounces spike, replies do not cancel follow-ups, or the send guard cannot prove a message has not already gone out.

What should you check before launching an automated campaign?

Here is the thing. Valid DNS will not protect you from a weak list, a sudden volume jump, or follow-ups that continue after a prospect replies. You need checks across the complete workflow, not only a green status inside the sending tool.

Use this pre-launch view before the first US prospect enters the sequence.

| Stage | Check before launch | Pause when | Fix or monitor | |---|---|---|---| | Domain and DNS | SPF, DKIM, DMARC, and From domain alignment | A record is missing, invalid, or misaligned | Correct DNS, reverify, then send test messages | | Mailbox pool | Every mailbox can send and receive normally | A mailbox has login, billing, connection, or reputation warnings | Remove it from rotation and test it separately | | Lead data | A small Apollo or Instantly lead-finder batch has passed checks for identity, fit, website, and mail-server status | The identity conflicts, the domain cannot receive mail, or the lead is blocklisted | Reverify or remove the lead before drafting | | Sending controls | Caps, spacing, send window, and deduplication work | Volume jumps, sends cluster, or duplicate risk appears | Freeze the affected queue before retrying | | Replies and bounces | Reply tracking, bounce capture, suppression, and follow-up cancellation work | Replies do not stop follow-ups or bounces remain queued | Stop the sequence and repair state handling | | Alerts | A person is notified when the system cannot decide | Failures are logged but nobody owns them | Assign an owner and a required response |

Is your domain authentication actually ready?

Start with SPF and DKIM. Google explicitly tells senders to set up SPF and DKIM authentication. Then publish DMARC so receiving providers can evaluate alignment and you can inspect authentication reports. Do not copy a DNS record from another domain and assume it fits. Verify the exact sending domain and selector used by each mailbox provider.

Alignment matters too. Yahoo advises senders to align the From domain with either the SPF domain or the DKIM domain. Check the headers from a real test message, not only the status shown by your outreach platform.

Before launch, send tests to multiple mailbox providers. Check the inbox, spam folder, authentication headers, reply threading, and opt-out handling. If authentication fails or replies disappear, I would not release the campaign.

How should you build and warm a mailbox pool?

One mailbox should not carry the growth target for the whole agency. Add capacity through a rotating pool, with a cap and state for each mailbox. New mailboxes should join gradually after DNS, sending, receiving, and inbox-placement checks pass. Warm-up is a readiness phase. It cannot rescue poor targeting or an abrupt increase in sending.

In our podcast outreach system, we built around a rotating pool of mailboxes. Each mailbox has 26 slots for new-prospect emails and 3 reserved for replies to people already in a conversation. New mailboxes are warmed in Instantly and checked for DNS and inbox placement before entering the pool. Cold emails go out between 11:00 and 15:00 US Eastern, the daily start time can shift by up to three hours, sends have random gaps, and weekend follow-ups wait until Monday.

Those figures describe our setup, not a universal formula. We accept fewer new prospects per mailbox so active conversations still have room. Protecting a reply is more valuable than squeezing another cold email into the queue.

For an agency targeting US prospects, start with a consistent recipient-friendly window and change it only after you have enough clean campaign data to compare. Do not change the window, list source, and copy at the same time.

Your mailbox record should show the domain, provider, warm-up state, daily cap, sends today, reserved capacity, latest warning, and active or paused state. Read that record immediately before every send.

How do you stop bad lead data before it reaches the sender?

Verification belongs before copy generation and campaign import. In our guest-booking workflow, small batches come from Apollo and Instantly's lead finder. A research agent checks each person against the guest brief: who they are, what they have done, whether they fit, and whether their email, website, and identity agree. Weak fits and mismatches are closed before a draft exists.

The workflow also checks whether the address's mail server can receive mail. Anyone who has refused, bounced, or asked not to be contacted goes onto a permanent blocklist and is never imported again.

For your agency, a US founder with a working email can still be a bad lead if the business cannot buy your service. Geography and a valid mailbox do not equal fit. Write the research agent's brief so it can reject that lead before copy is generated. If the brief cannot explain who counts as a real buyer, better automation will only make bad decisions faster. That is not an AI problem. That is a briefing problem.

What volume controls should the automation enforce?

Your automation should block unsafe sends, not merely log them. If you use n8n, see our production-safe n8n cold email automation workflow.

I would implement the sending path in this order:

1. Read the latest mailbox and pool state immediately before the provider call. Reject the send if either cap or pause state blocks it.
2. Record a durable send intent with an idempotency key. A retry must find that record instead of creating another message.
3. Release the provider call, save the result, and update the queue. If the result is uncertain, send it to review rather than retrying blindly.

Keep separate pause controls for a mailbox, domain, campaign, and data source. That lets you stop the damaged part without shutting down healthy conversations elsewhere.

In our setup, a launch is recorded before it happens, so a crash or restart cannot release the same email twice. The queue resumes from known state.

Change one part of the system at a time. Add a mailbox, change a lead source, or adjust a cap. If you change all three together, you will not know which change caused the warning.

Who is this checklist not for?

If you send at low volume from one mailbox, you may not need a rotating pool, durable send guard, and layered pause controls. Manual checks may be enough. This checklist is for teams scaling automated sending, where a retry, bad import, or disconnected inbox can affect an entire campaign before someone notices.

When should the bounce guard pause a campaign?

A bounce guard should act before a weekly report. It should close bounced leads, remove pending follow-ups, update campaign state, and pause the smallest unit that explains the problem. If the spike belongs to one imported list, stop that list. If it follows one mailbox or domain, isolate that sender. I would pause the whole campaign if the source is unclear.

OneAway recommends keeping bounce rate under 3% and spam complaints under 0.3%. Those are OneAway's suggested guardrails, not guarantees. A sharp change from your normal pattern can justify a pause before either level is reached.

When the guard fires, inspect the newest data source, verification result, receiving-provider pattern, DNS status, copy change, and sending mailbox. Reverify queued leads. Resume only after you can name the cause and show the fix in a test.

How do you protect replies and follow-ups?

Poll each sending inbox, classify replies, stop follow-ups immediately, and preserve the thread. Questions, out-of-office messages, assistants, refusals, and bounces need different states. Anything uncertain goes to a person.

The exit rule comes first: once a prospect replies, no scheduled follow-up may pass the send check. We rebuilt cold outreach for a Guwahati agency serving US clients around that discipline. The result was a 10.6% reply rate and no missed follow-ups.

If your automation overlooks a reply and sends another nudge, it damages a real conversation. That is why the reply check belongs inside the sending path, not in a report someone reads later.

Our guide to lead follow-up automation with clear stop and handoff rules shows how to keep prospects moving without sending the wrong next message.

Give every prospect a clear way to opt out, then suppress that address across future imports. Test the suppression path before launch. A person who opts out should not return through a new Apollo export or a different campaign.

What should you monitor every day and every week?

I care more about replies, bounces, complaints, disconnected mailboxes, stuck queues, and failed jobs than a flattering open-rate chart. Open rates are less reliable because of Apple’s Mail Privacy Protection, so they should not decide whether a mailbox is healthy.

Daily checks

• Mailboxes that are paused, disconnected, or close to their cap.
• Sudden changes in bounces or complaints by mailbox, campaign, provider, or lead source.
• Replies waiting for classification or human action.
• Follow-ups still queued after a reply, bounce, or opt-out.
• Failed jobs, uncertain provider responses, and duplicate-send warnings.

Weekly operating review

Review authentication, mailbox warnings, source quality, suppression, reply classifications, and failed alerts. Run inbox-placement tests, sample messages for broken variables, test the global pause, and confirm that the queue resumes without duplication.

Give each problem one owner and one action: keep, reduce, pause, reverify, or retire.

Build the controls before you scale

When something breaks, pause the smallest affected unit first, then widen the pause only if you cannot isolate the cause.

If you want this built around your domains, tools, lead sources, and sales process, tell Social Surge Media what you need and we will scope the custom automation.

---

About the author. Arindam Sikder, Founder, Social Surge Media. Builds AI automation systems for content, SEO, cold email and go-to-market, and writes about what works in practice.