10 Cold Email Mistakes That Kill Deliverability (the Ones We Actually Find in Audits)

1. Sending Too Much per Inbox

The written provider limits are 2,000 a day on Google Workspace and 10,000 recipients on Microsoft 365, and they are irrelevant. What governs cold email is the volume at which your engagement and complaint signals stop looking like a person's. That is about 15 cold sends per inbox per day on Google, 8 on Microsoft, 2 on Azure inboxes that make it up in count. Older guides, including the 2025 version of this page, said 50 to 100. Fleets run at that pace and are rebuilt every quarter. The fix is arithmetic: 500 a day is about 34 Google inboxes, and the inbox planner does the sum.

2. Cold Volume From a Domain That Never Warmed

A domain registered on Monday and sending 300 cold emails on Tuesday is the fastest burn there is. Receivers treat no history as a reason for suspicion, and the first week of a new domain's life decides how the next quarter goes. Warm for 14 to 21 days, rising from a handful a day to the working pace with real replies in the mix, or launch on prewarmed domains that already did it. The warmup timeline has the week-by-week ramp.

3. DMARC Missing, or DKIM Never Enabled

Since 2024, SPF, DKIM and DMARC are prerequisites at Google, Yahoo and Microsoft, not advantages. The two versions of this mistake we see most: a domain with SPF and nothing else, and a Microsoft 365 domain with the DKIM CNAMEs in DNS but DKIM never switched on in Exchange admin, so the key does not exist. Run the authentication checker, then read a delivered header; the authentication guide has the records to publish.

4. Tracking Through a Shared Domain

Sequencer default tracking rewrites your links through a domain shared with every other customer, so their complaints are your link reputation. Either set a custom tracking domain on your own sending domain or turn tracking off. We turned open tracking off fleet-wide and published what happened. The tracking domain checker shows what yours resolves to today.

5. Lists Nobody Verified

Bounces are the loudest signal you send. A list exported last quarter from a database and never verified bounces at 5% or more, and at that rate receivers stop reading the rest of your signals. Verify before the first send, re-verify anything older than 60 days, and treat catch-all addresses as their own problem. The verification waterfall keeps the cost down; the CSV cleaner strips role addresses and malformed rows before the verifier sees them.

6. A Link in the First Touch

The first message to a stranger with a link in it is the single strongest structural spam signal a cold email can carry, and it is also the one that costs replies even when it lands. Ask a question in plain text, earn the reply, send the link in the second message or after a response. The link timing article is position by position.

7. No Global Suppression List

An opt-out that only lives in one sequencer gets emailed again from the next one, and from the next client's campaign. One shared list of unsubscribes, hard bounces and hard noes across every domain, inbox and tool, checked before every send, is the difference between a complaint rate you control and one you do not. The suppression list guide builds it; the suppression checker runs a list against yours.

8. Bounces Ignored Until the Listing

Bounce rate is the one deliverability number you can read in your own tool every morning, and it predicts listings by days. Pause a domain at 2%, stop and investigate at 3%. Paste the bounce messages into the bounce analyzer to turn codes into causes: unknown users mean the list, deferrals mean pace, policy rejections mean reputation. The pause rules are the thresholds we use.

9. Nothing Watching the Fleet

Google Postmaster Tools is the only receiver-side view of your reputation that exists, it is free, and most fleets never connect it. Add every domain with Gmail volume, watch the spam rate and the domain reputation band weekly, and pair it with a weekly blacklist check and a domain expiry check, because an expired sending domain is a day of bounces nobody planned. The fleet monitoring article puts all three on one dashboard.

10. No Reserve for the Day It Burns

Something will burn. A domain gets listed, a tenant gets flagged, a client's list turns out to be a trap farm. The fleets that keep booking meetings that week are the ones with prewarmed capacity on standby equal to about a week of sending; the ones that do not start a 21-day warmup on the worst possible day. It is the third step of the pattern we run ourselves: prewarmed to launch, fresh to scale, prewarmed in reserve. The inboxing playbook has the full pattern, and if you want the reserve built for you, that is what we deliver, on Google, Microsoft 365 and Azure, within 48 to 72 hours.