Last updated: August 2026 · By the SMK Vietnam editorial team — a decade in SEO, content, and AI-assisted marketing for Vietnamese and foreign brands.
In Vietnamese marketing copy, the pronoun you choose is the brand voice. Use bạn for mass-market, youth, tech and D2C brands where equality and friendliness matter; use anh/chị for consultative selling, real estate, insurance, healthcare, B2B and any conversation where the reader is being personally advised; use quý khách for banking, aviation, luxury retail, official notices and legal or transactional messages where distance signals seriousness. The rule of thumb practitioners in Vietnam actually use: bạn = we are peers, anh/chị = we are talking to you as a person, quý khách = we are an institution addressing a customer. AI can draft in all three registers reliably — but only if you give it an explicit register instruction, because default AI Vietnamese drifts to a stiff, translated-sounding bạn.
Why register decides conversion in Vietnamese
English has one word for the reader: you. Vietnamese has a dozen, and every one of them encodes age, status, distance and warmth before the reader has processed a single benefit statement. A landing page that opens with Quý khách vui lòng để lại thông tin reads like a bank form. The same page opening with Bạn để lại thông tin nhé reads like a startup. Neither is wrong — but only one matches the product, and Vietnamese readers detect the mismatch instantly, the way an English reader would notice a hoodie brand writing “Dear Esteemed Patron.”
This is the single most common failure point we see when foreign brands localise into Vietnam. The translation is grammatically perfect and commercially dead, because the register was never specified — the translator or the AI model picked a default, and the default is almost always a flat, neutral bạn that suits nobody in particular.
The good news: register is one of the few localisation problems that AI handles genuinely well, because it is a rule-governed, patternable decision. What AI cannot do is choose for you. That is a brand decision, and this article is how to make it once, write it down, and then let the machine execute it consistently across a thousand assets. If you have not yet documented your voice, start with our guide to the complete AI marketing stack for Vietnam.
Figures above are directional estimates from our own client work, not published survey data — treat them as planning heuristics.
The four registers, and who each one is for
1. Bạn — the peer register
Neutral, friendly, age-blind. The default for consumer tech, apps, fashion, F&B, education, gaming, D2C and anything where the audience skews under 35. Pairs with mình or the brand name as the first person. Works everywhere on social, in-app and email. It is safe — which is also its weakness: over-used, it makes a premium brand sound generic.
Sounds like: “Bạn đã sẵn sàng chưa? Mình gửi bạn mã giảm 20% cho đơn đầu tiên.”
2. Anh/chị — the consultative register
Respectful without being cold. This is how a good Vietnamese salesperson talks: you are being advised by a competent junior. Essential for real estate, insurance, healthcare, wealth, B2B services, high-ticket home appliances and anything sold by consultation. Usually paired with em or chúng tôi as first person. The trap is gender and age — anh/chị written as a slash is fine in copy, but in a one-to-one message you should resolve it from CRM data.
Sounds like: “Em gửi anh/chị bảng so sánh 3 căn hộ phù hợp với ngân sách anh/chị đã chia sẻ.”
3. Quý khách — the institutional register
Formal, distant, authoritative. Banking, insurance policy documents, airlines, luxury retail, utilities, transactional notifications, terms and legal notices. First person is chúng tôi or the company name. It signals “this message is official and you should read it carefully” — which is exactly why it kills a playful campaign but rescues a payment-failure notice.
Sounds like: “Chúng tôi trân trọng thông báo tới quý khách về việc thay đổi biểu phí từ ngày 01/09.”
4. Cô/chú, con/cháu — the deferential register
Niche but powerful. Used when the audience is meaningfully older than the brand persona: health supplements for parents, elder care, traditional medicine, funeral services, rural agricultural products. Getting this right creates enormous warmth; getting it wrong — calling a 40-year-old cô — is an instant unsubscribe. Use it only when you are confident about the age profile.
Sounds like: “Cô chú để lại số điện thoại, các cháu sẽ gọi lại tư vấn miễn phí trong hôm nay.”
The industry matrix: pick once, apply everywhere
Below is the working matrix we hand to clients. It is not linguistics — it is a commercial default that you override deliberately, not accidentally.
“The register is not a style preference. It is a promise about the relationship — and Vietnamese readers hold you to it for the rest of the funnel.”
Five mistakes foreign brands make (and how AI fixes four of them)
- Register drifts mid-page: ad uses bạn, landing page uses quý khách, the sales SMS uses anh/chị.
- Global HQ approves an English tone doc that has no Vietnamese equivalent, so every freelancer improvises.
- Machine translation defaults to bạn even for a bank statement.
- Gendered anh/chị guessed wrongly in one-to-one Zalo messages.
- Formal register used in TikTok captions, killing reach.
- One written register rule per channel, stored in the team prompt library.
- A Vietnamese voice appendix attached to the global brand book.
- An explicit register line in every AI prompt, never relying on defaults.
- Resolve anh or chị from CRM gender/age; fall back to name-only greeting.
- An AI QA pass that flags register mismatches before publishing.
The one AI cannot fix is the second: somebody senior has to decide. Everything else is enforcement, and enforcement is exactly what a model with a clear instruction is good at. We covered the storage side of this in our Claude playbook for Vietnam, where project-level instructions hold a register across an entire campaign without re-prompting.
The 45-minute register audit
Paste your homepage, three top landing pages, five recent ads and two transactional emails into one document. Nothing else — this is a sample, not a corpus.
Ask the model to tag every second-person pronoun by asset and return a frequency table. You are looking for inconsistency, not correctness. Most brands discover two or three registers running at once.
Using the industry matrix above, write one line per channel: acquisition ads, website, in-app, email, SMS/Zalo, transactional, legal. A human with authority signs this off. It never changes without a new sign-off.
Turn the rule into a reusable system prompt block with two good and two bad examples per register. Store it where the whole team writes from — a shared project, not somebody’s laptop.
Add a final QA prompt that re-reads every draft and returns only the lines where the register breaks. Run it before publishing, not after.
Prompt patterns that hold register
Three patterns do almost all the work. Each assumes you are writing in Vietnamese and have already decided the register.
“Viết bằng tiếng Việt. Xưng hô: gọi người đọc là anh/chị, tự xưng là em. Không dùng bạn, không dùng quý khách. Giọng tư vấn, ngắn gọn, không hoa mỹ.” Two sentences, and it survives an entire batch of thirty product descriptions.
Ask for the same headline and first paragraph in all three registers, side by side, then read them aloud with a native speaker. This takes four minutes and settles arguments that otherwise consume a whole meeting. It is also the fastest way to convince a global stakeholder that the choice is not cosmetic.
“Đọc bản nháp dưới đây. Chỉ liệt kê các câu có xưng hô sai so với quy tắc: người đọc = bạn, thương hiệu = mình. Không viết lại, chỉ trích dẫn câu và nêu lỗi.” Returning only violations — not a rewrite — keeps the human editor in control and makes the output scannable in seconds.
If you run these at volume, put them in a shared library rather than pasting them each time. Our companion pieces on ChatGPT for marketing in Vietnam and Gemini in Google Workspace cover where each tool stores reusable instructions, and Perplexity for Vietnam research is useful when you need to check how competitors in your sector actually address their readers.
Key takeaways
- bạn = peers. anh/chị = personal advice. quý khách = institution to customer. cô/chú = deference to elders.
- Choose one register per channel and write it down; drift between assets costs more conversions than any single weak headline.
- AI never picks the right register by default — it defaults to a flat bạn. Always state the register explicitly in the prompt.
- Even inside one brand, transactional and legal messages usually shift up to quý khách while campaigns stay in bạn. That is correct, not inconsistent — as long as it is deliberate.
- Run a drift-detector prompt as the last step before publishing. It costs seconds and catches the errors native speakers notice instantly.
- For one-to-one messaging, resolve anh or chị from CRM data; if you do not know, greet by name rather than guess.
Related reading
- The Complete AI Marketing Stack for Vietnam — the pillar guide this article sits under.
- Claude AI for Marketing in Vietnam — storing register rules as project instructions.
- ChatGPT for Marketing in Vietnam — custom instructions and reusable prompt setup.
- Gemini in Google Workspace for Vietnam — enforcing voice inside Docs and Gmail.
- Perplexity AI for Vietnam Research — checking competitor register in your sector.
About SMK Vietnam
SMK Vietnam (smkvietnam.com) is a marketing hub serving Vietnamese companies and foreign brands from the US, Japan, Korea, Thailand and Europe marketing into Vietnam. We work on AI-assisted content, SEO and GEO/AEO visibility, localisation and marketing operations — with a bias toward systems a local team can actually run after we leave.
Not sure which register your brand should own?
We will run the 45-minute audit on your live assets and hand back a one-page register rule your team and your AI tools can both follow.
Talk to SMK VietnamFrequently asked questions
Is “bạn” ever unsafe to use in Vietnamese marketing?
It is rarely offensive, but it is often wrong for the situation. Using bạn in a bank notice, an insurance policy, a legal notice or a message to a customer clearly older than the brand persona reads as careless rather than friendly. Treat bạn as the safe default for marketing content aimed at a general or younger audience, and consciously upgrade to anh/chị or quý khách for consultative and institutional communication.
Can one brand use different registers on different channels?
Yes, and most mature brands do. A fintech may speak as bạn in TikTok campaigns and as quý khách in transaction alerts and terms of service. The requirement is that the split is documented per channel and applied consistently, so a reader never encounters two registers inside a single journey step — for example an ad and the landing page it points to.
How do I handle anh/chị when I do not know the customer’s gender?
In one-to-many copy, writing anh/chị with a slash is completely normal and expected. In one-to-one messages on Zalo, SMS or email, resolve gender from CRM data where you have it. Where you do not, greet by name — “Chào chị Lan” becomes simply “Chào Lan” or the more neutral anh/chị — and never guess, because a wrong guess is more damaging than a slightly formal greeting.
Does AI reliably keep the register I ask for?
Modern models hold a stated register well within a single output and reasonably well across a batch, provided the instruction is explicit and includes the forbidden alternatives. Drift appears in long documents and in follow-up turns where the instruction has scrolled out of focus. The practical fix is to store the register rule as a persistent project or system instruction rather than a chat message, and to run a short drift-detector prompt over the finished draft.
What should a foreign brand entering Vietnam decide first?
Decide the register before commissioning any Vietnamese copy at all. It changes headlines, CTAs, email subject lines and even visual tone. Add a one-page Vietnamese voice appendix to the global brand book covering the register per channel, the brand’s own first-person pronoun, and a short list of approved and banned phrasings. Doing this at the start avoids re-writing an entire launch package later.
Do register choices affect SEO or AI search visibility?
Indirectly but meaningfully. Register affects dwell time, bounce and conversion, which feed the engagement signals search engines observe. For AI answer engines it matters differently: a page written in a consistent, natural register is easier for a model to quote cleanly, while copy that lurches between formal and casual tends to be paraphrased or skipped. Consistency helps both.
Leave a Reply