Three different things get called deliverability and conflating them is why most diagnoses go wrong.
- Delivered, the receiving server accepted the message. It did not bounce. This tells you almost nothing.
- Inbox placement, it landed in the primary inbox rather than spam or promotions. This is what you actually care about.
- Opened, a human looked at it. Increasingly unmeasurable, because privacy proxies fire open pixels automatically.
A campaign can show 98% delivered and near-zero inbox placement. Every sending tool reports the first number prominently and the second not at all, because the first is flattering and the second is hard to measure.
The three records, and what they actually do
SPF, DKIM and DMARC are authentication, not reputation. They prove the mail came from where it claims. Setting them up correctly does not make you trusted, it makes you identifiable, which is a prerequisite for being trusted and also for being reliably blocked.
| Record | What it proves | What it does not do |
|---|---|---|
| SPF | This server is allowed to send for this domain | Say anything about content or reputation |
| DKIM | The message was not altered in transit | Vouch for the sender's behaviour |
| DMARC | What to do when SPF or DKIM fails | Improve placement on its own |
All three are table stakes. Configure them and you have removed a reason to be blocked; you have not given anyone a reason to trust you.
What actually drives placement
Engagement, mostly. Filters watch what recipients do: opens, replies, deletions without reading, and above all spam complaints. A domain that consistently produces replies keeps landing in inboxes. A domain that produces silence and the occasional complaint stops.
What Spamhaus thinks you are doing
Worth knowing before you spend money on this: Spamhaus, whose blocklists a large share of the internet consults before accepting mail, publishes its position on cold email, and the position is that unsolicited commercial bulk mail is spam. Not spam-adjacent. Spam.
Every vendor page in this category frames deliverability as a technical problem with a technical solution. The organisation that operates the filters frames it as a consent problem. Both descriptions are accurate about different things, and the gap between them is the actual risk you are managing.
Warmup, and what it is really for
Warmup services send mail between networks of accounts and mark it as read and replied. The stated purpose is establishing a positive sending history on a new domain. That is real, and it works for the specific case of a domain with no history at all.
What it cannot do is repair reputation while you are actively damaging it. Running warmup alongside a campaign generating complaints is applying and removing the same signal simultaneously, and the campaign is louder.
The setup that survives
- Send from a separate domain to your main one. A burned outbound domain must not take your company email with it.
- Authenticate all three records before the first send.
- Start low — tens of mails a day per mailbox, not hundreds, and increase over weeks.
- Verify the list. Bounces are the fastest way to signal you do not know who you are mailing.
- Make replying easy and one-click unsubscribing easier. An unsubscribe is vastly cheaper than a complaint.
- Watch reply rate, not open rate. It is the metric filters weight and the one you can still measure honestly.
If reply rate sits under about 2% on a clean, well-targeted list, the problem is upstream of sending. Stop optimising infrastructure and go back to the offer.