we deleted 151 mailboxes and it still wasn't the problem

a couple of weeks ago i sat down with my team and we deleted a hundred and fifty one mailboxes.

not paused. deleted. they'd been built wrong months earlier, they couldn't be activated, and they were squatting on names we needed. so we killed them and rebuilt from scratch. that same session we went from 268 live mailboxes to 296, and the target on the whiteboard is somewhere between five hundred and a thousand.

if that sounds like a lot of machinery for sending email, it is. domains, hosting, an smtp provider, the sending tool, spf and dkim and dmarc on every domain, warmup running for months before a single real message goes out, list cleaning on the way in.

an operator told me last week that his entire outbound runs out of his crm and a gmail account. about five hundred emails a week. it works right now. it will not keep working, and when it stops it takes his real company domain down with it, the one his customers reply to.

so yes, build the plumbing. that part is not optional and it never was.

now the number that keeps me honest.

i pulled a full audit of our own sending fleet. 121,811 emails sent. two bounces. two, out of a hundred and twenty one thousand, which is a rounding error against a rounding error, and it means every domain, every dns record, every warmed mailbox was doing its job.

those 121,811 sends produced 139 replies.

that's about one tenth of one percent. the benchmark people quote for this year is around three percent. we're roughly thirty times below it, on plumbing that is close to flawless.

i could have looked at those two bounces and told myself a story about how healthy we are.

here's what it means. perfect deliverability buys you the right to be ignored at scale. it gets the message into the inbox. it has no opinion at all about whether the message deserved to be opened.

and this is the trap, because deliverability is measurable and message quality is not. so teams spend six months on dns records and warmup curves, because those give you a green dashboard, and nobody wants to sit with the harder question, which is that the thing you're saying is not worth answering.

the fix is never more sends. more sends is how you turn a message nobody wants into a message nobody wants at a hundred thousand a month. the fix is changing the ask, or changing who you're asking, and then testing that with the same seriousness you gave the spf record.

we're rebuilding ours around a different opening, one where the person on the other end is being offered something rather than sold something. early numbers say it moves. i'll publish those too when i have enough of them to mean anything.

if you'd rather not spend a year learning this on your own domains, that's what the managed engagement covers. we build and run the sending stack and the copy, you keep every relationship it touches, flat fee, no guarantees attached.

but the honest version is that we paid for that lesson with a hundred and twenty one thousand sends and a hundred and thirty nine replies.

so before you buy another fifty mailboxes... would you answer your own email?