How to Set Up a Custom Domain Email Address That Stays Deliverable
Custom domain email deliverability rests on three things you control: authentication records in DNS, a sending reputation you build slowly, and a habit of watching what receivers tell you. Everything else, the mailbox software, the web interface, the vendor you happen to use this year, is replaceable. Get the fundamentals right and you can move between providers without your mail landing in spam.
How mail from your domain actually gets judged
When you send a message, the receiving server does not take your word for who you are. It looks at the envelope sender, the From: header, the connecting IP, and a set of DNS records published under your domain. It then decides, using rules you cannot see and thresholds that change, whether to accept, quarantine or reject.
Three mechanisms do most of the work. SPF lists the hosts allowed to send for your domain, and the receiver checks the connecting IP against that list. DKIM attaches a cryptographic signature to the message, and the receiver verifies it against a public key in DNS. DMARC ties the two together: it tells the receiver what to do when a message claiming to be from your domain fails SPF or DKIM alignment, and it tells you where to send reports. Modern receivers increasingly expect all three, and a domain with no DMARC record is treated with suspicion by default at several large providers.
There is a fourth thing that matters as much and is harder to configure: reputation. Receivers track the history of the IP addresses and domains that send to them. A new domain has no history, so its first messages are judged conservatively. A domain that suddenly sends a large volume of mail to addresses that mark it as spam will be throttled quickly. Reputation is earned in small increments and lost in large ones, which is why the technical setup is the easy part.
Publishing the records: SPF, DKIM and DMARC
Start with SPF. It is a single TXT record at the apex of your domain, and it is a list of mechanisms evaluated left to right. A typical record authorises your mail provider and nothing else:
example.com. 3600 IN TXT "v=spf1 include:mailprovider.example -all"
The include: mechanism pulls in another domain's SPF record, which is how providers authorise their own sending infrastructure. The -all at the end means "reject anything not listed". Use it once you are confident the list is complete; ~all is a softer failure that receivers may ignore. Keep the total number of DNS lookups under ten, because the SPF spec limits them and exceeding the limit is itself a permanent error. You can check your record with a lookup tool or by querying DNS directly:
dig +short TXT example.com
"v=spf1 include:mailprovider.example -all"
DKIM is configured by your sending service, which generates a key pair and gives you a public key to publish as a TXT record at a selector name such as selector1._domainkey.example.com. The private key stays with the sender, which signs each message with a DKIM-Signature: header. The receiver fetches the public key and verifies the hash. Rotate keys periodically and keep old selectors published until mail signed with them has aged out of queues.
DMARC is another TXT record, at _dmarc.example.com. Start in monitor mode and read the reports before you enforce anything:
_dmarc.example.com. 3600 IN TXT "v=DMARC1; p=none; rua=mailto:[email protected]; adkim=r; aspf=r"
The p=none policy asks receivers to take no action on failures but still send aggregate reports to the address in rua. Those reports are XML, ugly and voluminous, and they are the only way to find out which third parties are sending mail as your domain without your knowledge. Run in p=none long enough to see a clean picture, then move to p=quarantine and eventually p=reject. The adkim and aspf settings control whether alignment is relaxed or strict; relaxed is the usual starting point because it tolerates subdomain differences.
Custom domain email deliverability after the records are live
Once the records validate, the work shifts to sending behaviour. Warm a new domain gradually: send to people who expect to hear from you, keep volume low and steady at first, and avoid any sudden burst. A domain that goes from a handful of messages a day to thousands overnight looks like a compromised account, and receivers will treat it that way.
Watch the feedback loops. Many large receivers offer complaint feedback loops that report addresses that marked your mail as spam; sign up for them where available and honour unsubscribe requests immediately. Bounces matter too: remove hard bounces from your lists rather than retrying them, because repeatedly sending to addresses that no longer exist is one of the clearest signals of a careless sender.
Keep your DNS tidy. A dangling MX record pointing at a host you no longer use, or a stray SPF include: for a service you cancelled, is both a deliverability problem and a security one. Audit the records when you change providers, and remember that DNS changes take time to propagate, so plan migrations around that delay rather than discovering it mid cutover.
Finally, keep your sending infrastructure and your receiving infrastructure separate in your head. Your MX records determine where mail for your domain is delivered; your SPF, DKIM and DMARC records determine what you are allowed to send. They can live with different providers, and moving one does not require moving the other. That separation is what lets you change hosts without losing the reputation you have built, provided you keep the authentication records consistent across the move.
Next, pull your current DNS records and check them against what your sending provider says it needs, then run your domain through an SPF and DMARC validator and read the output carefully. If DMARC reports are already arriving, open one and see who is sending on your behalf. Fix the gaps before you tighten the policy, and set a reminder to review the records whenever a service you use changes its sending hosts.
