Sending from your own domain: SPF, DKIM, DMARC and MAIL FROM for SaaS founders
To send email from your own domain and have it delivered, you publish a few DNS records: DKIM keys that let receivers verify each message was signed for your domain, an SPF record and an MX record on a dedicated MAIL FROM subdomain so bounces come back to a name that is yours, and a DMARC record that tells receivers what to do when a message fails both checks. After that, deliverability depends on what you send and to whom: bounces and complaints are what damage a domain, and both have to be handled the moment they arrive.
This article explains each record in the terms a SaaS founder needs, what changed for bulk senders in 2024, and the mistakes that most often send good mail to spam.
Why does the From domain matter so much?
Because mailbox providers no longer judge a message only by its content. They judge the domain it claims to come from, and they keep a history of that domain.
When you send from you@yourproduct.com, a receiving server asks three questions. Was this message really authorised by yourproduct.com? Has it been altered on the way? And what does the history of yourproduct.com say about whether people want mail from it? The three records below answer the first two. Your sending behaviour answers the third, and it is the one that takes months to build and days to lose.
What does SPF actually check?
SPF (Sender Policy Framework) is a TXT record that lists the servers allowed to send mail for a domain. A receiver looks up the record and checks whether the server that delivered the message is on the list.
The detail most people miss: SPF checks the envelope sender, the address used during delivery and where bounces are returned (often called the Return-Path or MAIL FROM), not the From address the recipient sees. If your sending service uses its own envelope domain, SPF can pass for that domain while saying nothing at all about yours.
Two practical limits:
- One SPF record per name. Two TXT records starting with
v=spf1on the same name make SPF fail. Merge them. - Ten DNS lookups at most. Each
include:costs at least one lookup, and nested includes add up. A record that includes six services can break silently.
What is DKIM, and why is it usually three records?
DKIM (DomainKeys Identified Mail) adds a cryptographic signature to each message. The signature names a domain (the d= tag) and a selector, and the receiver fetches the matching public key from DNS at selector._domainkey.yourproduct.com. If the signature verifies, the receiver knows the message was signed by someone holding that domain's key and was not altered after signing.
Sending services commonly ask you to publish the key as a CNAME pointing to a key they host, sometimes three of them, so that they can rotate keys without asking you to edit DNS again. That is why a setup screen often lists three DKIM records rather than one.
DKIM is the more robust of the two checks. It survives forwarding in most cases, where SPF usually does not, because the forwarding server is not on your list.
What is a custom MAIL FROM domain for?
It makes SPF count for your domain.
By default the envelope sender belongs to the sending service, so SPF authenticates them and not you. A custom MAIL FROM is a subdomain of yours, typically something like mail.yourproduct.com, used as the envelope sender. You publish two records on it:
- an MX record, so that bounces have somewhere to go, and that somewhere is the sending service's feedback handling;
- an SPF TXT record on that subdomain, authorising the sending service.
Now SPF passes for mail.yourproduct.com, which is a subdomain of the domain in your From line. That relationship is what DMARC calls alignment.
What does DMARC add?
DMARC is a TXT record at _dmarc.yourproduct.com. It does two jobs.
First, it ties the checks to the domain the reader sees. A message passes DMARC when SPF or DKIM passes and the domain it authenticated is aligned with the From domain. In the default relaxed mode, a subdomain such as mail.yourproduct.com aligns with yourproduct.com. Without alignment, a message can pass SPF and DKIM for somebody else's domain and still be a forgery of yours.
Second, it states a policy for messages that fail: p=none (report only), p=quarantine (treat as suspicious) or p=reject. The rua= tag gives an address that receives aggregate reports, which show who is sending mail with your domain in the From line.
A sensible order is to start at p=none with a report address, read the reports for a few weeks to find every legitimate service that sends as you (your billing tool, your support desk, your own mailbox), fix their authentication, and only then tighten the policy. Jumping straight to p=reject is how a founder discovers that their invoices stopped arriving.
What did Gmail and Yahoo change in 2024?
Both announced requirements for bulk senders that took effect from February 2024. Google defines a bulk sender as one sending close to 5,000 messages or more a day to Gmail addresses; the rules are worth following well below that volume anyway. As published, the main points are:
- authenticate with both SPF and DKIM;
- publish a DMARC record (a
p=nonepolicy is accepted), with the From domain aligned with SPF or DKIM; - for marketing and subscribed messages, support one-click unsubscribe as described in RFC 8058, using the
List-UnsubscribeandList-Unsubscribe-Postheaders, and honour requests promptly (Google says within two days); - keep the spam rate reported in Google's Postmaster Tools below 0.3%, with Google advising to stay under 0.1%.
The providers update their guidance, so read their current sender pages before relying on any number here. The direction will not change: authenticated, aligned, easy to leave, rarely reported.
What breaks deliverability once the records are right?
DNS gets you to the door. What gets you through it is the reaction of the people you write to. Most damage comes from a short list of habits.
Sending to addresses that do not exist. A hard bounce (the mailbox does not exist, or the domain does not accept mail) should suppress the address immediately and for good. A soft bounce (mailbox full, server temporarily unavailable) can be retried a few times. Repeatedly sending to dead addresses tells providers your list is not maintained, and old lists are full of them.
Ignoring complaints. When a recipient presses "Report spam", some providers send a complaint back through a feedback loop. That contact should never receive marketing mail from you again. A complaint rate that looks small in absolute terms is large by provider standards: the thresholds above are fractions of a percent.
Volume that jumps. A new domain or a domain that has been quiet has no recent history. Going from nothing to several thousand messages in a day looks like a compromised account. Increase volume over a few weeks, starting with the people most likely to open and reply.
Mail nobody asked for. Bought lists, scraped lists and "everyone who ever signed up since 2019" produce bounces and complaints together. Write to people who have a reason to expect you.
Hiding the exit. An unsubscribe link that is hard to find becomes a spam report, because the spam button is always easy to find.
Should marketing mail use a separate subdomain?
It is a reasonable choice, and a trade-off rather than a rule. Sending campaigns from news.yourproduct.com and transactional mail from yourproduct.com separates their reputations to a degree, so a badly received campaign is less likely to delay a password reset. The cost is a From address your readers recognise slightly less. For a small SaaS writing to its own sign-ups at modest volume, sending from the main domain with clean lists is often fine; for anything cold, separate it.
Where Canopus fits
Canopus is an autonomous growth manager for SaaS: it writes your strategy from Stripe and your product, drafts the campaigns, sends them from your own domain once you approve them, and tells you what each one earned. You do not connect an email tool. You add your sending domain, and the settings screen shows the records to publish: the DKIM CNAMEs, the MX and SPF records for the MAIL FROM subdomain, and a DMARC record. Canopus checks that they resolve before anything can be sent.
After that, campaigns are sent from your verified domain with RFC 8058 one-click unsubscribe and a daily cap. A permanent bounce or a complaint stops that contact in every campaign at once and flags the contact in the CRM. Bounce and complaint rates are watched per domain over seven days, and sending for a domain is paused when they cross a threshold, before the damage spreads.
The comparison with an email platform shows how this differs from connecting a sending tool yourself, and the pricing page gives the monthly price excluding VAT.