Rules that only travel one way

Three levels, one direction

Rules are set at the account, at a brand and at a campaign, and everything wider is inherited by everything narrower. A level below may add rules of its own and may raise one it inherited from advisory to blocking. That is the whole of it: it may not delete an inherited rule, drop it to advisory, reword it, or change which kinds of content it reaches.

If a brand team can quietly drop a house rule to advisory, the house rule was never a rule. It was a default.

When a brand tightens, the severity is the only part that becomes the brand’s. The definition stays the account’s — the statement, the pattern, the replacement — so a brand cannot rewrite what a house rule says while appearing to obey it. Reword the rule at the account and every brand that made it stricter is quoting the new sentence the next time it loads, still at the severity it chose.

Refused at the control, not at the save

The weaker severity cannot be chosen on an inherited blocking rule and there is no remove button on an inherited rule at all, each with the reason drawn beside the control. That is deliberately earlier than the save: somebody who believes they removed a rule and did not is worse off than somebody who was told no, because they find out at export with whoever was waiting for the post watching.

A weakening that reaches the data by some other route is dropped rather than honoured. There is a test named for the way round this that people try, which is to save twice: load the rules, drop one to advisory, save, load again, save again. The rule is still blocking.

What is not there yet is the account’s side of it. A weakening that gets past the control is discarded quietly — the function that would name it back, rule by rule, exists and is tested, and nothing in the product calls it. So the rule holds and nobody is told an attempt was made.

Deleting a house rule, and what it costs

A tightening is stored as an override that says what it is, so the register never credits the account’s wording to whoever merely made it stricter. Delete the rule at the account and those overrides are dropped rather than promoted into rules the brands now own — otherwise a brand would go on blocking exports on the authority of a house rule that no longer exists, citing a document the account has withdrawn.

That costs something. Deleting a house rule removes enforcement at exactly the brands that cared enough to strengthen it, the severity they chose is not kept anywhere, and re-adding the rule starts every brand back at the account’s.

The case this does not catch

An override stored before it carried the flag that marks it as one is indistinguishable from a rule the brand wrote itself. Those rows stay, and stay wrong. They could be guessed at from the document they cite, and guessing would delete rules brands actually authored in order to tidy up ones they did not. A flag is evidence and a resemblance is not.

The full table of what each level may and may not do is on rule inheritance.