Gates, verdicts, and why NO-DATA blocks

Skills in this toolkit render verdicts: SEND-CLEAR, POLICY-BLOCKED, LAUNCH-BLOCKED, a tracking grade, a scale-or-cut call. They all share one grammar. Learn it once and every skill’s output reads the same way.

The four row values

Every individual check resolves to exactly one of four things:

Value	Means	What it must show you
PASS	Checked, and it holds	The number or the quote, never just the word “present”
FAIL	Checked, and it does not hold	The number or the quote
NO-DATA	Could not be checked. Access, input or permission is missing	Exactly what is missing, and how to get it
N-A	Genuinely does not apply to this account	The evidence that establishes it does not apply

The rule that matters most

NO-DATA blocks exactly like FAIL blocks

A check that could not run never reads as a check that passed. A week where the dataset was unreachable does not come back looking like a clean week.

This sounds obvious and is the single most common way audit tools lie to people. The natural failure is quiet: a gate has no slot for “I could not evaluate this,” so it picks the nearest confident answer, and you get a green report about an account nobody actually looked at.

So: a clear verdict requires zero failures, zero NO-DATA, and zero unevidenced gaps. Satisfying the stated condition for “clear” is necessary and not sufficient.

Two consequences worth internalizing

Claiming something is absent requires having looked. A skill may report “no exclusions found” only if it read the exclusions. If it could never see them, the honest answer is NO-DATA, not a clean bill of health.

Verdicts are computed from what was evaluated, not from what was found. A check that ran over nothing has not run. That is why you will see denominators like 34 / 51 evaluated next to a verdict rather than a bare pass, and why “zero problems found” on a run that examined zero things is reported as NO-DATA.

N-A needs evidence

“No meaningful EU traffic” backed by a geo distribution is N-A. The same sentence backed by nothing is NO-DATA. The most common real answer from an account owner is “I don’t know,” and that is NO-DATA too.

How evidence inside a deliverable is graded

Separately from gate verdicts, individual claims in a deliverable carry an evidence tier:

Tier	Means
RECEIPTED	Traced to an id in a verified bank
UNBANKED-PROVISIONAL	A real observed quote or number that is not banked yet, anchored to its raw source
OPERATOR-HYPOTHESIS	Asserted by you, or authored as craft

The middle tier is the one that matters. Without it, a real customer quote you have not formally banked gets either rejected or relabelled as invented, and the second one misrepresents a real person’s words as copywriting.

What this means for you day to day

• Read the NO-DATA rows first. They tell you what to go connect.
• A blocked launch or a restricted send is the system working, not an obstacle. Both exist because the alternative costs more.
• If you disagree with a verdict, argue with the evidence column, which is always shown.
• Don’t route around a gate. They are the reason a green report from this toolkit means something.

Related

Where numbers come from covers what happens when a required number cannot be sourced at all.