YundaDeskA buyer’s first message is often just “Hi”, an order number, or a local-language version of “is anyone there?” Your first reply does not need to be clever. It needs to show two things fast: you understand the buyer, and you know what to do next.
Many cross-border teams already have a greeting, but every country and every channel gets the same English line. A steadier setup is a small set of multilingual greetings selected by the CRM language field. When the buyer continues, AI support follows their language; when the case is complex or high risk, it hands off to a human.
Start with the rule: greetings are not a translation contest
A multilingual greeting is not about translating one sentence into ten languages word for word. It is about getting buyers into the right support flow quickly. Hold three rules:
Multilingual Greeting Templates That Match the Buyer's Language: the service baseline to plan around
- Say what you can help with: orders, shipping, sizing, product details and after-sales issues.
- Do not ask twice: if the CRM, order record or social identity already has the data, do not make the buyer repeat it.
- Keep the human exit visible: AI catches repetitive questions first, while refunds, compensation and price changes always need human approval.
A useful opener is not “Welcome to our store.” It is closer to “I am here. Send your order number or question, and I can help with shipping, sizing or after-sales support.” That turns politeness into usable information.
Use CRM language fields to choose the opener
Do not make agents guess language from the channel. A WhatsApp buyer does not always prefer English. An Instagram DM may arrive in Spanish, Thai or Arabic. A better setup is to match the opener from the customer record:
| Priority | Language source | How to use it |
|---|---|---|
| 1 | Previous conversation language | Reuse the language that worked last time |
| 2 | CRM language field | Best for known buyers and captured leads |
| 3 | Country, timezone, browser signal | Helpful, but only as supporting context |
| 4 | Customer’s first message | Let AI detect and switch for first-time visitors |
YundaDesk’s cross-border CRM includes country, language, timezone and social IDs as default fields, with automatic identity merging. If the same buyer comes from the website widget today and WhatsApp tomorrow, agents and AI still see one customer profile.
Multilingual greeting template: cover your top markets first
The set below is not final copy. It is a multilingual greeting template your support lead can adapt: I am here, here is what I can help with, and here is what you can send next.
| Language | Template |
|---|---|
| English | Hi, I am here to help. Send your order number or question, and I can check shipping, sizing, product details or after-sales options. |
| Spanish | Hola, estoy aquí para ayudarte. Envíame tu número de pedido o tu pregunta, y puedo revisar el envío, la talla, el producto o las opciones de posventa. |
| Portuguese | Oi, estou aqui para ajudar. Envie o número do pedido ou sua pergunta, e posso verificar envio, tamanho, detalhes do produto ou pós-venda. |
| French | Bonjour, je suis là pour vous aider. Envoyez votre numéro de commande ou votre question, et je peux vérifier la livraison, la taille, le produit ou le service après-vente. |
| German | Hallo, ich helfe gern weiter. Sende deine Bestellnummer oder Frage, dann prüfe ich Versand, Größe, Produktdetails oder After-Sales-Optionen. |
| Japanese | Hello, I can help with your order. Send your order number or question, and I can check shipping, size, product details or after-sales support. |
| Thai | Hello, send your order number or question and I can help check delivery, sizing, product details or after-sales support. |
You do not need 30 languages on day one. Check order countries, ad markets and social inquiry sources, then build the first five to eight markets properly. A maintained short list beats a long list nobody trusts. If channels are still scattered, fix that with an omnichannel inbox first.
Keep the self-introduction short
In a support conversation, buyers do not need a brand history. They need to know whether this is official, whether you can solve the issue, and whether a human will step in when needed. Keep the self-introduction to one line:
| Moment | Suggested line |
|---|---|
| Website widget | I am the YundaDesk AI agent. I can handle common questions first and bring in a teammate when human judgment is needed. |
| Social DM | This is the brand support team. I can check orders and product questions first; refunds or compensation will be confirmed by a human teammate. |
| Email auto-reply | We received your message. AI will prepare an answer from the knowledge base first, and a human will continue if the case is complex. |
AI support should not pretend to be a person or promise to handle everything. It catches repetitive questions first; humans back up the moments that need judgment. For the operating model, see AI answers first, humans back up.
Turn small talk into usable information
After the greeting, do not stop at “How may I help you?” It is too broad. Guide by intent so the buyer gives information you can process:
| Intent | Prompt |
|---|---|
| Track an order | Send your order number and I will check the latest shipping status. |
| Change an address | If the order has not shipped yet, send the new address and I will check whether it can still be changed. |
| Pick a size | Share your height, weight and usual size, and I will compare them with the product size chart. |
| Ask about a product | Send the product link or screenshot, and I will check material, specs and fit for your use case. |
| After-sales issue | Send the order number, a short description and photos. Anything requiring approval will go to a human. |
AI can collect after-sales details, explain policy and calm the conversation. But refunds, compensation and price changes should not run automatically. They need approval and an audit trail.
Adjust by channel: do not paste the same line everywhere
Language changes, and channel tone should change too:
- Website widget: keep it short and focus on products, sizing and shipping.
- WhatsApp / Telegram / LINE / Zalo: make it conversational, but avoid sending several messages in a row.
- Instagram / TikTok / Messenger: carry the product context first, then ask for order or sizing details.
- Email: confirm receipt, set expectations and list what information the buyer can add.
- Custom API: pass page, product or cart context into the conversation so the buyer does not repeat what the system already knows.
YundaDesk brings website widget, custom API, email, WhatsApp, Telegram, Messenger, Instagram, TikTok, LINE, WeChat, VKontakte, Zalo and YouTube into one workspace. Templates, customer records and the knowledge base do not need to live in channel silos.
Make the template smarter over time
The point is not to write a perfect template once. Improve it as real conversations show where buyers hesitate or misunderstand:
- The support lead confirms the base greeting for each language
- QA checks whether AI openers match the customer’s language field
- Agents tag greetings that buyers repeatedly misunderstand
- When an agent corrects AI, the system creates a learning suggestion
- An owner accepts the suggestion before the new version goes live
- Every change stays traceable, testable and revertible
That is what YundaDesk means by “gets smarter over time.” When AI misses an answer, when an agent fills the gap, or when an agent corrects AI, the system creates learning suggestions you confirm. Only after the owner approves them do they become skills, knowledge or customer memory. Learning never takes effect automatically.
Appendix: pre-launch checklist
- Does the CRM have a language field, and can greeting rules read it?
- Does first-message language detection switch templates for new customers?
- Does every target market have a greeting, self-introduction, shipping prompt and after-sales prompt?
- Do refund, compensation, complaint and review-threat messages trigger a human handoff?
- When the customer asks for a person, can the conversation switch to an agent with full context preserved?
Multilingual greetings are not about looking international. They reduce buyer hesitation, agent repetition and AI mistakes. Pick the opener from the CRM language field, let AI continue in the buyer’s language, and route high-risk moments to humans. That is steadier than sending every customer the same English greeting.