Do You Need a Separate Domain for Cold Email?

The case for sending cold outreach from a separate domain is usually made in terms of deliverability, and that framing undersells it. The strongest argument has nothing to do with inbox placement.

Your primary domain carries your invoices. It carries password resets, contract notifications, support replies, and the address every existing customer uses to reach you. If a cold campaign damages that domain's reputation badly enough, the failure is not that your outreach stops working. It is that a customer does not receive their receipt, a prospect never gets the proposal they asked for, and a colleague's reply to a live deal lands in a spam folder.

Cold outreach is the highest-variance mail any company sends. It goes to people who did not ask for it, from a list you cannot fully verify, and some proportion of recipients will mark it as spam no matter how careful you are. Running that traffic through the same domain as your billing is a decision to put your most critical mail behind your least predictable mail.

Why a subdomain is weaker than it looks

The obvious compromise is a subdomain: keep company.com for real business, send outreach from mail.company.com. It feels like isolation with none of the overhead of a new domain.

It is not, quite. Receivers evaluate reputation at the organisational domain level as well as the exact hostname. Sustained abuse from a subdomain can and does affect how the parent domain is treated, because the alternative — letting anyone insulate a domain by inventing a new label in front of it — would make subdomains a trivial evasion route. Filters closed that door a long time ago.

A subdomain is a perfectly reasonable structure for transactional mail, product notifications, or anything you control tightly and send predictably. For cold outreach, where the entire point of separation is to contain a risk you cannot fully control, it provides less containment than the diagram suggests.

A separately registered domain is genuinely separate. That is what you are paying the registration fee for.

Choosing a domain that does not look suspicious

A separate sending domain only helps if it reads as legitimate. Receivers are actively suspicious of newly-registered domains that resemble established ones, because that is precisely the shape of a phishing attempt.

A few things reliably help. Choose a name obviously connected to your company — getacme.com, acmehq.com, try-acme.com are all fine; a string of hyphens or an unusual TLD is not. Put a real website at it, even a single page that explains who you are and links to the main site, so anyone who checks finds a company rather than a parked domain. Register it well before you need it: domain age is a genuine input, and a domain that has existed for three months when you start sending is treated differently from one registered on Tuesday.

Then warm it. A separate domain does not inherit your primary domain's reputation, which is the entire point, and it also means it starts with nothing. Every argument for gradual ramping applies to it in full.

How many domains you actually need

This is where advice online goes badly wrong. There is a widely-repeated pattern of buying ten domains, attaching several mailboxes to each, and rotating sends across all of them to reach a volume no single domain could support.

The logic is backwards. Domains do not create sending capacity; reputation creates sending capacity, and reputation is earned per domain, slowly. Spreading a volume you have not earned across ten cold domains does not give you ten times the headroom. It gives you ten domains with no history, each sending unfamiliar mail to strangers, in a coordinated pattern that looks exactly like the abuse this technique was invented to disguise.

The defensible reason to run more than one sending domain is that your volume has genuinely outgrown what one warmed domain carries comfortably, and you want to distribute load across identities you have each warmed properly. That is a consequence of scale, arrived at after the fact. It is not a shortcut to scale.

For most teams the honest answer is one sending domain, warmed properly, doing all the work — with a second added later if and when volume demands it. Every additional domain is another identity to authenticate, warm, monitor and eventually explain to somebody. They are not free, and the cost is ongoing.

A practical setup

A structure that works for the large majority of outbound programmes looks like this. Keep the primary domain for everything that matters and never send cold volume from it. Register one lookalike domain, put a real page on it, and publish SPF, DKIM and DMARC for it exactly as you would for the primary. Warm it gradually over several weeks before it carries a real campaign. Watch its bounce and complaint rates as its own numbers, separate from anything else you send.

If that domain reaches its comfortable ceiling and the programme is still growing, add a second and repeat the process. If it does not, you are done — and you have one identity to look after instead of ten.

RepMail registers and verifies your sending domain through AWS SES, publishes the authentication records for it, and applies a progressive daily ramp to a new sender automatically rather than leaving warm-up as something you have to remember. Because the entitlement and the domain are workspace-level, adding teammates does not quietly multiply the number of identities sending against that one domain's reputation.