What a brand-voice check cannot see

It is worth being precise about what this class of tool does, because the gap is where expectations break.

What it catches

Violations your own guide can be quoted against: banned terms where the guide says never/avoid, preferred-term misses where it says use X instead of Y, emoji and exclamation policies, all-caps, hedge density, passive constructions, sentence-length limits. These are checkable against the pasted text, so they are checked against the pasted text — deterministically, with quoted evidence for every finding.

What it cannot catch

• Copy that is on-brand and still bad. A draft can obey every rule in the guide and fail to make a single argument. Obedience is not persuasiveness.
• Missing voice. A check compares what is written against what the guide says. It cannot report a positioning claim that was never made, or a section nobody wrote.
• Stale intent. The brand may have moved on and the guide may not have been updated; copy that faithfully matches an outdated guide is "compliant" in a way that no longer means anything.
• Context the paste does not contain. Sarcasm, cultural register, and channel expectations live outside the two texts you pasted. Tone is multi-dimensional (see Nielsen Norman Group's dimensions of tone of voice); a rule check covers the dimensions that rules can encode, not the ones only a reader can feel.

Why the boundary matters

A tool that claimed to verify brand quality would be claiming something it cannot deliver, and teams would find that out at the worst moment. Stating the limit up front means the check gets trusted for the things it genuinely knows — and the human reviewer stays exactly where a voice check should leave them: in charge of the copy.