Priced by storage, not by headcount

Disk and bandwidth. That is close to the whole list. A mailbox is a directory of messages and an index; the cost of running it scales with how much mail is in it and how often it moves, and barely at all with whose name is on the door.

So a price built on headcount is measuring something the server does not feel. It is a good proxy for value in software where a person is load — a seat in a CRM, a participant in a call — and a poor one here, where the busiest human on the team and the address that receives a monthly invoice cost the same to keep.

Pricing on storage puts the meter where the money goes. It has an honest consequence, which is that the customer with ten years of attachments pays more than the customer who deletes things, and that is the correct direction for the difference to run.

Storage is one figure for the account, not an allowance per mailbox. Ten gigabytes are shared across every mailbox on every domain, and a mailbox holding 40 KB is not wasting an allocation it was never given. Extra storage is bought in 5 GB blocks at $0.40/GB/yr and applies to the pool, so buying more never means choosing which mailbox gets it.

A mailbox may also carry its own cap, and this is the part worth reading twice: a cap is a ceiling, never a reservation. It can only subtract. Capping one mailbox at 2 GB stops that mailbox at 2 GB; it does not set 2 GB aside for it, and it does not protect it if the pool fills up. Both gates apply, and either one can stop a delivery.

That makes the counter-intuitive case explicit: if the account pool is full, mail to a mailbox sitting comfortably under its own cap queues anyway. The cap did not buy space the account does not own. Over-subscription — the sum of the caps exceeding the pool — is fine and is the entire point of pooling.

Inbound mail queues rather than bouncing, and it is delivered once there is room. That is measured behaviour against the mail server we run, not an assumption: an over-quota delivery is accepted and held as a temporary failure, and raising the account’s storage releases the held messages on the queue’s own next attempt with no intervention.

The dashboard shows the percentage of the pool in use, because a raw byte figure tells you nothing about whether you are near the line. Adding storage is self-serve and prorated against the time left on the subscription.

Measured usage is used for display and for that one enforcement flip, and it is never multiplied by a price. The only billed quantity is the number of blocks you chose. That distinction is the difference between a plan and a metered bill, and it is why there is no invoice you cannot predict.

Five people on Google Workspace Business Starter is $420 a year, and each of them gets 30 GB that cannot be lent to anyone else. Five people here is $9 with 10 GB between them. Buying up to 150 GB of pooled storage — well past what five ordinary mailboxes hold — puts the account at $69 a year, still a sixth of the Workspace bill.

The comparison flips when the archive gets genuinely large. Migadu meters messages per day instead and treats storage as soft, so a 200 GB archive with light sending costs less there than here. That is a real answer for a real buyer and the Migadu comparison says so at length.

Every competitor figure on this site was read off that vendor's own public pricing page on 12 August 2026, for five people, for a year.