How to Write a Chatbot Persona and Brand Voice Guide
A practical template for your chatbot persona and brand voice guide: naming, formality per language, do's and don'ts, emoji rules, example rewrites and handling angry customers.

في هذه المقالة
Two chatbots answer the same question: "Do you ship to Norway?"
One says: "Yes, shipping to Norway is available. Please refer to our shipping page for details." The other says: "Yes, we ship to Norway. Delivery takes 4 to 6 working days, and import fees are shown at checkout." Both are correct. Only one sounds like a company that wants your order.
That difference comes from a chatbot persona: a short written guide that defines how your assistant sounds, what it calls itself, how formal it is in each language, and how it behaves when things go wrong. Without one, an AI assistant falls back on a polite, generic voice that could belong to any brand.
Key takeaways
- A chatbot persona is a one- to two-page guide covering identity, tone, formality, do's and don'ts, and example rewrites.
- Decide formality per language and write it down, because English hides choices that other languages force.
- Whatever the persona, the assistant should always be clear that it's an AI and make a person easy to reach.
The chatbot persona template
Copy this structure into a document and fill in each line:
- Purpose: what the assistant is for, e.g. "Answer product and order questions for our online store and pass complex cases to the team."
- Name: a name, or simply "[Brand] assistant"
- Three personality traits, each with a limit: "friendly but not chatty," "confident but not pushy," "expert but not technical"
- Formality per language (see below)
- Vocabulary: words you use and words you avoid
- Reply length: e.g. "Two to four sentences unless the customer asks for detail"
- Emoji policy
- Difficult situations: angry customers, complaints, out-of-scope requests
- AI disclosure and handoff: how it introduces itself and how it offers a person
Good persona documents fit on one or two pages. Any longer and nobody, human or AI, follows them consistently.
Should your chatbot have a name?
Both choices work. A named persona ("Hi, I'm Juno, the [Brand] assistant") feels friendlier and gives customers something to refer to. A plain label ("Hi, I'm the [Brand] assistant") is simpler and avoids any suggestion that customers are chatting with a person.
The rule that matters: never imply the assistant is human. No fake typing delays, no invented backstory, no human first name paired with a job title. The EU AI Act transparency obligations we flagged in our 2026 support trends point the same way.
Set chatbot tone of voice per language
English uses "you" for everyone. Many European languages make you choose:
| Language | Informal | Formal | Things to weigh |
|---|---|---|---|
| German | du | Sie | Informal suits young consumer brands; formal is safer for B2B, finance and health |
| French | tu | vous | "Vous" is the usual choice for businesses |
| Spanish | tú | usted | Informal is common for consumer brands in Spain; norms vary by country |
| Dutch | je | u | "Je" is widely used by consumer brands |
| Polish | ty | Pan / Pani | Formal address is the safer default in business |
Write the decision for each language into the guide, and make sure your assistant and your agents use the same register. A customer who goes from "du" with the bot to "Sie" with an agent notices. For more on language quality, see our guide to multilingual customer support.
Chatbot brand voice: do's, don'ts and rewrites
Do's and don'ts
| Do | Don't |
|---|---|
| Answer first, then add context | Open with "Great question" |
| Reuse the customer's own words | Use internal jargon or SKU codes |
| Give specifics: dates, prices, times | Say "soon" without a timeframe |
| Apologize once, sincerely | Apologize in every message |
| Offer a clear next step | End every reply with "Anything else?" |
| Admit what you don't know | Guess to sound helpful |
Example rewrites: generic vs. on-brand
Here's how one hypothetical outdoor gear store might rewrite stock replies:
| Generic reply | On-brand reply |
|---|---|
| "Your inquiry has been received." | "Got it. Here's what happens next." |
| "Unfortunately, that item is out of stock." | "That jacket is sold out in medium right now. The large runs close to size if you'd like to compare." |
| "Please contact customer support." | "I'll pass this to our team. They'll reply right here in this chat." |
| "We apologize for any inconvenience." | "Sorry, that's not the experience we want you to have. Let's fix it." |
| "Is there anything else I can help you with?" | "Anything else before your trip?" |
Put five to ten rewrites like these into your guide. Examples steer an AI assistant's tone far better than adjectives.
Emoji policy and angry customers
Emojis. Decide once: never, sparingly or freely. A sensible middle ground is "only if the customer uses them first, at most one per message, never in complaints." Some emojis read differently across cultures, so keep them out of anything serious.
Angry customers. Personality should step back when someone is upset. Write these rules into the guide:
- Drop jokes, exclamation marks and emojis.
- Name the specific problem: "Your order arrived two days late, and it was a gift for today."
- Don't argue with the customer's account of what happened.
- Offer a person early. Our guide to chatbot-to-human handoff has wording you can reuse.
Putting your persona into Conviro
In Conviro, the tone and boundaries you write become instructions for your AI assistant: how it sounds, what it won't discuss, and when it passes a conversation to your team in the Live Inbox. It answers from your knowledge base, so the persona shapes the voice while your content supplies the facts. If you manage several brands, the new multi-brand workspace lets you run separate branded assistants, each with its own knowledge base and analytics.
Frequently Asked Questions
Should a chatbot have a human name?
It can have a name, but it shouldn't pretend to be human. A name like "Juno, the [Brand] assistant" is fine if the greeting makes clear it's an AI. Avoid human first names combined with job titles or personal stories, which mislead customers and damage trust once they realize.
Should chatbots use emojis?
It depends on your brand and audience. Casual consumer brands can use them sparingly; finance, health and B2B brands usually shouldn't. A safe rule is to mirror the customer: use an emoji only if they do, never more than one per message, and never in complaints or sensitive conversations.
How do I make my chatbot sound less robotic?
Answer first instead of opening with filler, use specific details, vary sentence length and cut stock phrases like "We apologize for any inconvenience." Give the assistant example rewrites of generic replies in your brand voice. Concrete examples shape tone far better than labels like "friendly" or "professional."
Next steps
Draft your persona with the nine-point template, add five rewrites taken from your own transcripts, and test the assistant against them. Start free with Conviro and put your voice guide to work.
نشرت من قبل
كونفيرو
مشاركة الرؤى حول دعم العملاء المدعوم بالذكاء الاصطناعي وأتمتة روبوتات الدردشة واستراتيجيات نمو SaaS.
ضع كتاب اللعب موضع التنفيذ
حوّل معرفتك بالدعم إلى إجابات أسرع
استكشف مساحة عمل دعم الذكاء الاصطناعي في Conviro، ثم ابدأ مجانًا عندما يناسب سير العمل فريقك.
مقالات ذات صلة
المزيد من المشاركات
Guides & Tutorials
Instagram DM Automation: A Beginner's Guide for Businesses
6 دقيقة قراءة
اقرأ المقال
Guides & Tutorials
Sending Images and PDFs in Chat: Brochures, Price Lists, and Menus on Demand
7 دقيقة قراءة
اقرأ المقال
Guides & Tutorials
AI Chatbot for WooCommerce: What It Can Do and How to Set It Up
7 دقيقة قراءة
اقرأ المقال