Last updated: November 2026 · By Khanh Nguyen, marketing lead at SMK Vietnam — a decade in SEO, content and AI-assisted marketing for Vietnamese and foreign brands
After a decade in SEO, content, and AI-assisted marketing in Vietnam, I can tell you where most margin quietly disappears: not in ad spend, but in promotions nobody modelled. A 12.12 voucher stacked on a platform voucher, stacked on a livestream-only price, stacked on free shipping — and a “record sales day” that lost money on every order. AI is very good at fixing exactly this problem, and very bad at fixing it alone.
Use AI for pricing and promotion strategy in Vietnam in four jobs: (1) monitor competitor prices and vouchers across Shopee, Lazada and TikTok Shop every week, (2) model the true margin of every stacked discount before a mega-sale, (3) generate and test promotion mechanics and Vietnamese offer copy, and (4) read post-campaign data to separate incremental sales from sales you would have made anyway. Keep one thing human: the price floor. AI recommends, a person with P&L responsibility approves.
Why is pricing and promotion in Vietnam different?
Vietnamese shoppers are among the most promotion-literate in Southeast Asia. They compare the same SKU across three marketplaces, wait for double-day sales (9.9, 10.10, 11.11, 12.12), collect platform and shop vouchers, and ask in Zalo or Messenger whether there is a “better price”. Promotions are no longer a tactic; for many categories they are the calendar. That makes discount depth a strategic decision, and most teams still set it in a spreadsheet the night before the campaign.
Figures are directional planning estimates, not audited benchmarks. Validate against your own data.
Where does AI actually help with pricing and promotions?
Six jobs earn their place. None of them require an enterprise pricing engine; a general assistant such as Claude or ChatGPT, a spreadsheet, and clean exports get you most of the value.
Competitor price monitoring
Paste weekly screenshots or exports of competitor listings. AI normalises pack sizes, flags price moves over 5%, and spots who is running which voucher.
Stacked-discount margin modelling
Give AI your cost, platform fees, shipping subsidy and every voucher layer. It builds the worst-case basket and shows contribution margin per order.
Promotion mechanic design
Ask for ten mechanics at the same cost: bundle, gift-with-purchase, tiered voucher, buy-2-save, member price. Value-added offers usually protect price perception better than a flat 30% off.
Vietnamese offer copy
Generate offer framing variants — “Mua 2 tặng 1” vs “Giảm 33%” are mathematically close and psychologically different. Test them, do not guess.
Demand and elasticity reads
Feed 12–18 months of orders with price and promo flags. AI estimates how volume responds to each discount depth and where an extra 5% stops paying.
Post-campaign truth-telling
Compare the two weeks before, during and after. AI separates incremental sales from pull-forward and cannibalisation of full-price SKUs.
For marketplace-specific tactics, pair this with our AI for Shopee & Lazada playbook; for the calendar itself, see AI for seasonal campaigns in Vietnam.
“A promotion you cannot explain on one line of a P&L is not a promotion. It is a donation.”
— what I tell every team before 11.11
AI-assisted pricing vs gut-feel discounting: what changes?
What AI does well
Models every voucher combination in minutes, not days
Reads competitor price moves consistently every week
Suggests non-price mechanics a tired team forgets
Writes a plain-language post-mortem leadership will read
Where it still fails
It does not know your real costs unless you provide them
Elasticity estimates on thin data look precise and are not
It cannot see distributor and dealer reactions to online prices
It will not check promotion registration or advertising rules for you
That last point matters. Promotions in Vietnam are regulated — certain programmes must be notified or registered with the authorities, and “original price” claims have to be genuine. Treat AI output as a draft and have your legal or compliance contact confirm the mechanic before launch. Foreign brands should be especially careful: an offer that is routine in Bangkok or Los Angeles may need paperwork here.
What does a 5-step AI pricing workflow look like before a mega-sale?
This is the workflow we run about three weeks before a double-day sale. Total effort is roughly one working day spread across the team.
Finance and marketing agree a minimum contribution margin per SKU after platform fees, shipping and returns. Write it down. This number is not negotiable by AI or by the livestream host.
Collect current prices and visible vouchers for your top 20 SKUs and three competitors each. Have AI build a comparison table and summarise positioning. Perplexity helps with public pricing research; marketplace data you collect manually or by export.
Conservative, base, aggressive. For each: discount depth, mechanic, expected uplift range, worst-case stacked basket, margin per order. Ask the model to show its arithmetic and check five rows by hand.
Generate Vietnamese offer copy in three framings, then run a small paid test or a Zalo broadcast split a week before. See our A/B testing guide for sample sizes.
Export orders, give AI the before/during/after windows, and ask one question: how much of this was incremental? Save the answer where next quarter’s planner will find it.
What mistakes do teams make with AI-driven pricing?
| Mistake | What happens | Fix |
|---|---|---|
| Asking AI for “the best price” | A confident number with no basis | Give costs, fees and history; ask for ranges and assumptions |
| Ignoring voucher stacking | Negative margin on the best-selling basket | Always model the worst-case stacked order |
| Same discount on every channel | Dealers and offline retailers complain | Differentiate by bundle or gift, not headline price |
| Permanent promotion | The sale price becomes the real price | Cap promo days per SKU per quarter |
| Measuring only revenue | “Record day” that lost money | Report contribution margin and incrementality |
| Uploading sensitive cost data carelessly | Confidential margins in the wrong tool | Use business-tier tools; anonymise SKUs where possible |
If you sell through loyalty programmes, the cheapest discount is usually the one you only show to members — our loyalty and personalization playbook covers targeted offers. And to prove the whole exercise pays, tie it to the framework in measuring AI marketing ROI.
Key takeaways
- In Vietnam, promotions are the calendar — so discount depth is a strategy decision, not a last-minute one.
- AI’s highest-value pricing job is modelling stacked discounts and worst-case margin before a mega-sale.
- Value-added mechanics (bundles, gifts, member prices) usually protect price perception better than deeper flat discounts.
- Always measure incrementality and contribution margin, not just campaign revenue.
- The price floor and legal sign-off stay human. AI recommends; a P&L owner approves.
About SMK Vietnam. SMK Vietnam (smkvietnam.com) is a marketing hub that helps Vietnamese companies and foreign brands from the US, Japan, Thailand, Korea and Europe plan and run marketing in Vietnam, with a practical focus on SEO, content and AI-assisted workflows.
Frequently asked questions
Can AI set prices automatically for my Shopee or Lazada store?
It can recommend prices, but we advise against fully automatic repricing for most Vietnamese SMEs. Use AI to monitor competitors and model margin, then have a person approve changes against a written price floor.
How much data do I need for AI to estimate price elasticity?
As a working rule, 12 to 18 months of order data with clear price and promotion flags, and several distinct price points per SKU. With less, treat any elasticity figure as a rough direction rather than a forecast.
Which AI tools are best for promotion planning in Vietnam?
A general assistant such as Claude or ChatGPT for modelling and copy, Gemini in Google Sheets for spreadsheet work, and Perplexity for public market research cover most needs. Dedicated pricing software only pays off at several hundred SKUs or more.
How do I stop mega-sale discounts from destroying margin?
Set a contribution-margin floor per SKU, model the worst-case stacked voucher basket before launch, prefer bundles and gifts over deeper flat discounts, and cap the number of promotion days per SKU each quarter.
Do promotions in Vietnam need legal approval?
Some promotional programmes must be notified to or registered with the authorities, and reference-price claims must be genuine. Rules change, so confirm each mechanic with a qualified legal or compliance adviser; AI output is not legal advice.
Is it safe to upload cost and margin data to AI tools?
Use business or enterprise plans with clear data-handling terms, anonymise SKU names where you can, and never include customer personal data in pricing files. Agree the rule internally before the first upload.
Planning 12.12 or Tết promotions?
We help Vietnamese and foreign brands build promotion plans that grow sales without giving away the margin.
Talk to SMK Vietnam
Leave a Reply