Most companies treat inbox placement as weather. It was good last quarter. It is worse this month. Someone files a ticket with the ESP. A specialist sends a PDF. The week continues as if the inbox were a channel they still owned.
It is not weather. It is operations. Authentication, list hygiene, complaint handling, the shape of the send graph, and the discipline to stop a stream that is poisoning the domain — that is a system. If nobody holds it, the domain decays in public, slowly, while the dashboard still reports opens.
Opens are a rumour. Placement is a fact, and even placement is only as honest as the seeds and the panels you trust. The operating question is simpler than the industry pretends. Did the mail we intended to send arrive where we intended it to arrive, to people who still want it, without teaching Gmail that we are a nuisance.
The craft starts before copy. SPF, DKIM, DMARC, a sending domain that is not a costume, a return path that is watched. Alignment is not a one-time IT project. It is a configuration that marketing will break the first time someone “just needs a new subdomain for a blast.”
Then the list. A list is not an asset because it is large. It is an asset because it is recent, permissioned, and still in a relationship with the company. Reactivation programmes that hammer three-year-old addresses are not growth. They are a tax on everyone else in the file.
The send graph is the part most teams refuse to draw. How much mail, from which IPs and domains, at what cadence, to which cohorts. A product launch that doubles volume overnight is not ambition. It is a reputation event. Throttle is not timidity. It is how grown programmes survive a quarter.
Complaint rate and unsubscribe are operating numbers. They belong in the weekly, next to revenue, not in a specialist’s appendix. If the VP cannot see them without a guide, the programme is not being run. It is being decorated.
Tools do not hold this. Salesforce Marketing Cloud, Braze, Klaviyo, Iterable — they will send whatever you ask, beautifully, into a hole. Fluency in the platform is necessary. It is not sufficient. The platform will not stop you. A person has to.
A retained practice holds deliverability the same way it holds a calendar. Diagnose: where placement actually is, which streams are toxic, which authentication is theatre. Build: the domains, the suppression, the warming plan, the stop rules. Run: a weekly look at the graph, not a quarterly panic. Measure: placement, complaint, unsub, time-to-launch. Hand back: a playbook the next owner can run without a séance.
The first thirty days are usually ugly in a useful way. You will find a forgotten journey still firing. You will find a sales sequence colliding with the brand programme. You will find a seed list that has not been touched since the last ESP migration. None of that is a scandal. It is the work. Pretending it is already fine is how domains die politely.
Warming is not a ritual for new IPs only. After a merge, after a re-platform, after a season of discount abuse, the domain needs a diet. Volume comes back when the relationship comes back. There is no coupon that repairs reputation.
Legal and brand are part of the system. A proof that takes five days is a deliverability problem in disguise: the send happens in a scramble, at the wrong hour, to the wrong overlap of lists. Time-to-launch is an inbox metric. Treat it as one.
What to stop doing is usually clearer than what to start. Stop buying lists. Stop “winback” that is actually harassment. Stop letting every product manager mint a journey. Stop reporting opens as if they were revenue. Stop assuming dedicated IPs are a personality.
Handover is the test. If the practice leaves and placement is a mystery again, you rented a specialist. If the team can see the graph, name the kill rules, and keep authentication from drifting, you installed an operating system. That is the only kind of deliverability work worth paying for.
MBX holds this as email work, not as a side quest. The inbox is a channel. Channels have owners. Owners have weeks. Write when the domain is already serious enough that weather is no longer an acceptable theory.
A note on ESP migrations, because they are where programmes pretend to start over. They do not. You bring the list, the complaint history, the forgotten journeys, and the habit of sending at 9 a.m. because that is when the proof finally cleared. The new platform will make all of that faster. Faster is not healthier. The first month after a migration should be a diet and a map, not a celebration send.
Seed tests are a flashlight, not the sun. Use them. Do not manage the programme to them. Panels disagree. Apple’s privacy features have already taught anyone paying attention that a single number will lie. Hold a basket: placement on the domains that matter to your file, complaint, unsub, and a commercial number the business already uses. When they move together, you know something. When only the seed moves, you know you overfit a flashlight.
Camille Edith Baudouin, MBX Consulting