Illingworth says 60 inboxes need $10 of domains, not $200

Richard Illingworth says the standard practice of buying a fresh domain for every two or three inboxes is paying for safety that does not exist. He says the belief driving it, that subdomains on one root all burn together, was tested on his own setup and the deliverability came out the same either way.

His arithmetic is the concrete part. For 60 inboxes, the separate-domain approach needs about 20 domains, which at $10 each is roughly $200 spent on domains alone. The subdomain approach needs one root domain, so $10 covers the same 60 inboxes. That is a twentieth of the cost for what he says is identical placement.

The second half of his argument is about what happens when something goes wrong, which it will. Illingworth's position is that every setup burns at some point regardless of what you spent, and the question is what a burn costs you. On a subdomain tree, he says a flagged branch gets cut off and the rest keep sending, which is why he thinks you can push more volume without treating each burn as a crisis.

What the post does not contain

No send volumes, no placement rates, no inbox-level numbers, no test period. Illingworth writes that "we tested it, and it doesn't make any sense" but does not publish the test. That makes this his read on his own infrastructure rather than a dataset, and it is not a number you can price a client rebuild against.

Richard Illingworth

@RCIllingworth

X

we tested it, and it doesn't make any sense

It is also not new ground for him. He has made the subdomain case before with at least one figure attached, saying subdomains held at 1.8M sends when the root domain stayed silent. The cost framing here is the new part, and it is aimed at the spend rather than the sending.

For people sending, the thing to weigh is not the $190 saved on a 60-inbox setup, which is noise next to inbox and tooling costs. It is concentration. Everything Illingworth describes depends on a flagged branch being cuttable without dragging the root and its siblings down with it, and that is exactly the assumption the separate-domain crowd refuses to make. He says his testing settles it. He has not shown the numbers that would settle it for anyone else's setup, so if you move, move a slice of volume and watch placement before you move the rest.