Multi-domain sending infrastructure
The main job was keeping cold email volume off the brand domain. If the brand domain gets flagged, invoices and client email go to spam along with the outbound.
- Role
- Architecture and ongoing operation
- Registrars
- GoDaddy, Namecheap, Porkbun
- Platforms
- Instantly, Smartlead, SMTP.com
- Held
- Placement at volume, brand domain intact
How it goes wrong
The sequence is almost always identical. A team decides to scale outbound. They send from the company domain, because that’s the domain they have. Volume climbs, complaint and bounce rates climb behind it, and a few weeks later the domain’s reputation is damaged.
The part that gets underestimated is the effect on everything else. Your invoices start landing in spam, and you find out from a client who says, slightly annoyed, that they never received it. Password resets, support replies and the contract you sent Tuesday all degrade at once, and none of it is obviously connected to the outbound experiment somebody started last month.
Reputation builds slowly and collapses fast. Every decision below exists to make that asymmetry survivable.
The architecture
Keep the brand domain out of it
Cold outreach never goes out on the domain the business runs on. Sending domains get bought separately, deliberately spread across several registrars so one registrar-level policy change can’t take the entire program down in an afternoon.
Warm before you send
A domain with no history that starts sending hundreds of messages a day looks exactly like a spammer, because at that moment there is nothing to distinguish it from one. Each domain and mailbox goes through scheduled warm-up, building volume gradually and generating real engagement before a campaign touches it.
Align the records
SPF, DKIM and DMARC configured and aligned for each domain. Records that exist but don’t align are worse than missing ones: they pass your checklist and fail at the inbox, so you go looking for the problem everywhere except the place it is.
Cap it, then watch it
A few mailboxes per domain, each with a conservative daily ceiling, volume spread across all of them. Reputation, bounce rate and placement get watched continuously so a degrading domain comes out of rotation and rests before it becomes a problem.
The argument I have with nearly every client
Warm-up is measured in weeks, and you are going to want to skip it. I understand why: you have a number this quarter and a rack of domains sitting idle. I hold the line anyway, because the alternative is rebuilding in six weeks on worse infrastructure than we started with, having spent the intervening time teaching Google that your domain sends unwanted mail.
- More surface to manage. Many domains and mailboxes means more configuration, more monitoring and more cost than one domain. That overhead is what buys the isolation.
- Capacity deliberately left on the table. On any given day the ceilings sit below what the domains could probably handle. I’d rather leave headroom than find the limit by crossing it.
The one rule I don’t bend
The domain your invoices go out on never sends cold email. Everything else on this page is negotiable depending on your risk appetite and your timeline. That one isn’t, and it’s the cheapest insurance in the entire stack.