Docs, brand voice
Custom Rules
Standing instructions the AI follows on every reply — sign-off lines, competitor restrictions, stylistic constraints. General verticals only; regulated industries (healthcare, legal) ignore this field entirely.
Why it matters
Most brand voice settings shape who you are — soul, tone, offerings. Custom Rules shape what the AI must always do or never do when it replies on your behalf. A sign-off that makes every response feel like it came from a real team. A hard rule against naming competitors. A stylistic constraint you can’t achieve with a tone preset alone. This is the escape hatch for standing instructions that don’t fit anywhere else.
Most owners leave this blank and get excellent results. Fill it in only when you have a specific, recurring requirement that the other fields cannot handle.
What to write
Plain sentences. One rule per line works well. Keep each rule short and unambiguous.
| Example rule | What it does |
|---|---|
Always sign off as -The Team. | Every reply ends with a consistent close. |
Never mention competitors by name. | Prevents any competitor shoutout, even when a reviewer brings one up. |
Keep a dry, understated tone. | Layers a stylistic modifier on top of the selected tone preset. |
Do not promise specific response times or service windows. | Blocks operational commitments the AI has no business making. |
2000 character ceiling. Keep it focused — a rules list longer than a few lines risks conflicting instructions and diminishing returns.
How the AI uses it
- Every reply gets the rules applied, regardless of rating, review length, or content.
- Rules layer on top of tone, soul, and offerings. They do not replace them.
- Restraints always win. If a custom rule conflicts with a built-in system restraint (no medical advice, no legal advice, no operational promises), the built-in restraint takes precedence. Custom rules cannot override the platform’s core guardrails.
- Regulated industries (healthcare, legal) — this field is hidden and ignored entirely. Practice-area and clinical instructions belong in the Brand Soul, not here.
Common mistakes
- Contradicting the tone preset. If your preset is “Warm & Approachable” and a custom rule says “Be cold and formal”, the AI gets conflicting signals. Pick one.
- Operational promises. “Always offer a 10% discount to unhappy reviewers” — this is exactly the kind of commitment the AI should never make in a public reply. The system restraint will override it.
- Rules that belong in Brand Soul. “Always mention that we’ve been in business since 1987” belongs in your identity prose, not here. Custom Rules is for behavioral constraints, not identity detail.
- Overstuffing. Four tightly written rules outperform twelve vague ones.
Cross-vertical examples
Restaurant — Always sign off as -The Pit Crew. Never mention competitors.
SaaS / software — Keep replies under four sentences. Do not promise feature timelines or roadmap items.
General contractor — Never quote prices or give estimates in a reply.
Sign off as -Mike and the crew.
See it in action
Our second visit this month. Food is consistently excellent and the room has a great vibe. Highly recommend.
Dana, thank you so much! Consistently excellent is exactly what we aim for, and we're really glad the atmosphere came through too. Hope to see you again soon.
Dana, two visits in a month means the world to us. Glad the food and the room are both landing right. See you on visit three. - The Team
The "before" reply is warm but generic. Adding a sign-off rule gives every response a consistent, recognizable close without any other change to tone or content.