TL;DR: Cold email deliverability comes down to a few decisions made before the first send: separate sending domains so your main domain never sends, many inboxes at low volume each, real warm-up, and attention to which mail provider the recipient uses. In our own sending, reply rates differed sharply between Google Workspace and Microsoft 365 recipients.

What is cold email deliverability?

Deliverability is whether your email reaches the recipient's inbox rather than the spam folder or a bounce. It depends on your sending domain's reputation, how your inboxes behave, the quality of your list and where the recipient's mail is hosted. Good copy cannot rescue an email that never arrives.

It is worth separating two failures that look alike. A bounce means the message was refused. Spam placement means it was accepted and then filed where nobody looks. Both show up as silence, and they have different causes. Bounces point mostly to the list, which we cover in how a cold email list gets built so it fits. Placement points to the sending setup, which is the subject here.

Because so many things feed into placement, the useful habit is to diagnose in a fixed order. Check the sending setup first, then the list, then the copy. Skipping straight to rewriting the email is the most common mistake, since it feels like the part you control, and it is often not the part that is broken.

Why use separate sending domains?

Cold email carries reputation risk, so it should never leave from your main domain. We set up separate sending domains for cold email, and your main domain, the one on your website and invoices, never sends one. If a sending domain is damaged, the business is not.

The logic is containment. Mailbox providers score a domain by how its mail is received. If cold sends carry your main domain and recipients ignore or report them, that score follows your normal business email too, and a client's invoice could land in spam. Separate domains keep the experiment away from the thing you cannot afford to damage.

Each client has their own set. Every outbound client of ours has their own sending domains, inboxes, workspace and do-not-contact list, so one account's behavior never affects another's. An opt-out goes on that list and is never emailed again, which is both courteous and good for reputation.

How do inboxes and per-inbox volume work?

Spread sending across many inboxes and keep each one's volume low. A single inbox sending a lot looks like a bulk sender to a mailbox provider. Many inboxes each sending a little look like what they are: people writing a small number of emails a day.

In our own sending, no inbox sends more than 11 cold emails a day. Our fleet runs about 275 inboxes at 11 sends a day and 280 at 2 a day on the lower-reputation tag, about 3,585 sends a day across roughly 935 inboxes. The total is large, but no single inbox is. Total volume comes from the number of inboxes, not from pushing any one of them.

A practical consequence is that a narrow market needs fewer inboxes than a broad one. Sending is planned around the list in front of us, and nothing is pushed harder than it should be to hit a date.

Low volume per inbox also leaves room. An inbox running well below a ceiling can absorb a normal bad day, such as a run of bounces, without tipping into trouble. An inbox run at its limit every day has no such margin.

What does warm-up do?

Warm-up builds a sending history before an inbox contacts a real prospect. New inboxes with no history look suspicious to a mailbox provider, so they send and receive a small, rising amount of mail first. In our service every inbox warms during the 14 days before launch.

Warm-up is not a switch you flip once. It is a measured signal, and we read it. In a comparison on the same sending tool, native Microsoft 365 and Google Workspace mailboxes ran a 7-day warm-up health score of 99, against 80 for our reseller-provisioned fleet. We now only send from inboxes with a miss rate at or under 40% and a 7-day health at or over 90.

That rule means a weak inbox does not send a cold email, even when it would be convenient. Nothing weak goes live. Holding the line on that costs a little volume and protects the rest of the fleet, which is the right trade.

Underneath all of this sits technical authentication. Every sending domain needs the standard records that tell a receiving server the mail is genuinely from you, usually called SPF, DKIM and DMARC. They are routine to set up and easy to get wrong, and a mistake there damages placement from the first send. We set them up on every domain as part of the build, so it is not something you have to check.

The email itself matters too. Plain text written like a note from one person, with no tracking links or attachments in the first message, sits better with mailbox providers than a designed newsletter. A first email also does not need a link or a bare domain. If someone replies, you can share one then.

Does the recipient's mail provider change the result?

Yes, a great deal. In our own sending across 16,570 cold emails, the reply rate was 0.02% to Google Workspace inboxes, 0.80% to Microsoft 365 and 0.89% to other hosts. That is a large gap, and it shapes how we plan lists.

The raw counts are worth stating: 2 replies on 10,528 sends to Google Workspace, 42 on 5,256 to Microsoft 365 and 7 on 786 to other hosts. This is reply rate by the recipient's provider, measured across all of our campaigns, so it describes our sending specifically, not email in general. We do not claim to know every reason behind it. We treat it as something to plan around.

What we do with it:

The practical lesson is that a list is not just a role and a size. Where its mail is hosted changes what you should expect, and a good plan accounts for it.

None of this is a reason to expect a number in advance. The reply gap says where the odds are different, and that is all. It is a reason to measure by provider and to be honest about what a given list can produce.

How do you keep deliverability healthy week to week?

Deliverability is maintained, not set once. Every week we read placement, bounce and reply data, remove anything that bounced or opted out, and hold back inboxes that slip. When replies are slow, we check the sending first, then the list, then the copy.

A weekly deliverability check covers a short list:

  1. Inbox health and miss rate, per inbox
  2. Bounce rate on what went out
  3. Replies, read by recipient mail provider
  4. Opt-outs added to the do-not-contact list
  5. Any inbox below the sending line

This is one part of what a done-for-you team does, and the full process is in our guide to done-for-you cold email. If you are comparing it with running this yourself, see done-for-you versus in-house or an SDR.

If you want the infrastructure handled and your main domain kept clear, our outbound service includes it. To talk through your market, book a discovery call.