Brand Voice & Multi-Language Replies
Written By Chad McGuire (Sparrow Intel)
Overview
The previous lesson covered how Chirp AI drafts a response. This one covers how it sounds — your team's voice, and the way it carries across languages.
Brand voice is the single biggest lever you have on review-reply quality. A team that takes 15 minutes to tune their voice settings ends up rewriting 70% fewer drafts.
Where brand voice lives
Open Review Settings to find:
Review Scale — 5-Star or 10-Point (covered in Admin Setup Checklist)
Review Reply Settings — your brand voice configuration
The Review Reply Settings card is what we're focused on here.
Two modes: presets and custom
You can configure brand voice in one of two ways.
Preset guidelines
Select checkboxes for built-in patterns. Common presets:
You can pick any combination. Most teams start with all three on, then tune from there.
Custom prompt
If presets don't capture your voice, write a custom prompt that tells Chirp exactly how to sound. Examples:
Reply in a warm, conversational tone. Sign every response with "— The team at [Brand]." Don't use exclamation points more than once per response. If the guest mentioned a problem, acknowledge it directly without making excuses.
Match the energy of the review. Short reviews get short replies. Don't apologize for things outside our control (weather, neighborhood noise). Always thank the guest for taking the time to write.
Custom prompts compose with the presets — you can use both. If a preset and your prompt conflict, the prompt wins.
Writing a good custom prompt
A few patterns we've seen work:
Lead with the rule, not the rationale. "Don't use exclamation points" beats "Try to avoid being too enthusiastic."
Call out anti-patterns explicitly. "Don't apologize for the weather" prevents a recurring miss.
Don't over-prescribe. Chirp does best with 3–6 clear rules. A 30-line prompt usually means you're trying to encode every property's quirks in the voice settings — the more rules you pile on, the less Chirp can prioritize the ones that matter.
Multi-language replies
When a guest reviews in a non-English language, Chirp AI:
Detects the language on the incoming review
Stores the original and an English translation so your team can read it
Drafts the response in the guest's native language
Applies your brand voice in that language
A response Chirp drafts in French will follow the same tone rules as one in English. Your custom prompt doesn't need to be translated — Chirp interprets the intent and applies it across languages.
If you'd rather respond in English to a non-English review, use the language selector in the composer to switch. Many teams reply in the reviewer's language for Airbnb and Booking.com (where guests are typically reading from one country) and reply in English on Google (where local search visibility matters).
Which languages does Chirp handle?
Chirp AI replies in any major language used by your guests — the same language coverage as Conversations. The most common are English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Russian, Mandarin, and Japanese, but the list is long.
If you spot a draft in a language where the tone or grammar feels off, let us know via in-product chat. Real examples are what we use to tune.
Brand voice and Auto-Respond
When Chirp AI sends a review response automatically (via a Rule action), it uses the same brand voice settings as when you trigger AI responses manually. So getting your voice right pays off twice — once for every reply you send by hand, and once for every reply a rule sends without you.
Tune brand voice before you turn on auto-respond rules. A bad voice setting amplifies fast when rules are running.
Up next
You've completed the Take-Off phase — you can now find, read, reply, and trust the AI suggestion in your team's voice.
Next is Triage: Star, Archive, Snooze & Bulk Actions — the lightweight workflow actions you'll use dozens of times a day.