I once watched a team spend a full week on one cold email. Sharp subject line, tight copy, clear ask. Then they sent it from their main company domain, got flagged as spam within two days, and took their invoices and client replies down with it. The email was never the problem. The setup underneath it was.

This is more common than you'd think. As the lemlist deliverability team likes to put it, most emails that land in spam were not written badly, they came from domains that were never set up properly (more on that here). Without the right foundation, Gmail and Outlook have no reason to trust you, so they don't.

By the end of this, you'll know how to set up your domains, mailboxes, and warm-up so your cold email lands in the inbox, and your main domain stays clean.

Cold email deliverability rests on three legs: your infrastructure, your warm-up routine, and your content. Everyone loves to argue about content. The opener, the CTA, the perfect subject line. I get it, that's the fun part. But if the infrastructure underneath is broken, none of it ships. So we start at the bottom.

Infrastructure means your domains, your mailboxes, and how you authenticate and ramp them up. Here's the order I set it up in:

This is the mistake I see most. Your main domain runs your invoices, your client threads, your whole business. If it gets burned (flagged as spam by providers), all of that breaks with it, not just your outreach.

So buy a separate domain just for outbound, and point it at your main website. If your outbound domain redirects nowhere real, that's a red flag for providers too.

A lookalike domain is a variation of your main name, used only for sending. Some patterns help you, others quietly hurt you.

Patterns that work:

Patterns to avoid:

Before you buy, check the domain's history. Search "[domain name] blacklist" on Google to see if it was flagged before. And when you pick where to host it, your provider matters more than people expect, so it's worth choosing a domain host built for outreach.

Patience is not optional here. Do not send any cold email in the first four weeks after buying a domain. This gives blacklist checks time to clear and lets the domain settle.

Same logic for a new mailbox. Send nothing in the first week. Put it in an answer-only mode that replies to existing threads, then warm it up for four to five weeks before any cold outreach. Warm-up is just the mailbox sending and receiving normal-looking mail so it builds a sender reputation before you scale. lemlist automates that whole process, and it also flags problems early instead of after you've torched a domain.

This comes down to two things: how many leads you have and how fast you want to reach them. The target is around 40 emails per day per mailbox max, and the safe range is closer to 25 to 35.

Two quick examples:

The rule underneath: the more you spread your volume across domains and mailboxes, the lower each one's limit can be, and the safer your overall reputation stays. If setting all of this up by hand sounds like a weekend you'd rather not lose, lemlist can provision the domains and mailboxes for you and handle the DNS.

Most business email runs on Google Workspace or Microsoft 365, split fairly evenly between the two. So don't stack all your mailboxes on one provider.

Spreading across both does two things. It protects you if one provider has an outage or a policy change, and it gives you more sender-recipient combinations when your leads sit on different providers.

This part I'll hand to Maxime, who owns deliverability at lemlist. Teams always ask him the same thing: what's the right way to ramp up with lemwarm before sending with lemlist? His answer is a 4-phase system, and it's the ramp-up he swears by. The whole point is that your real sending volume never goes over 40 emails a day per mailbox.

You can set the limit and the increment directly in lemwarm's setup, so none of this is manual.

Always keep a backup. Roughly 10% of your setup should sit idle and ready. 10 domains, keep 1 spare. 100 mailboxes, keep 10.

The reason is simple. If a mailbox or domain takes a hit mid-campaign, you swap in a backup the same day instead of scrambling to fix the original while your sequence stalls.

Infrastructure isn't a one-time setup, it's a system you keep running: separate domains, smart lookalikes, patient aging and warm-up, the right mailbox count for your volume, provider diversity, and backups on standby. Get this right first, because no clever subject line can save a domain that was built wrong from the start.