TL;DR
- A Penang-based indie fashion brand (composite name: "Threadline KL") doubled monthly online sales from RM38,000 to RM82,000 in 6 months using a RM480/month AI stack.
- Shopee SKU count went from 80 to 280 (4x more listings shipped) — same single owner-operator.
- Abandoned cart recovery improved from 2% to 9% using Klaviyo AI sequences.
- Meta Advantage+ AI cut Facebook ad CPM from RM22 to RM14 (-36%) while maintaining sales.
- The total monthly AI spend (RM480) is roughly 0.6% of the new monthly revenue (RM82,000).
[IMAGE 1 — Featured: Malaysian fashion brand product photography]
Intro
This case study covers how a Penang-based indie fashion brand selling on Shopee, Lazada, and their own Shopify store doubled monthly online sales in 6 months — from RM38,000 to RM82,000 — using a RM480/month AI stack. The owner runs the entire operation solo with one part-time content assistant. No new hires. Same marketing channels. The change was AI taking over the listing generation, customer service, abandoned cart recovery, and ad management work that was bottlenecking growth. Here's exactly what changed and what other Malaysian e-commerce sellers can replicate.
The Business: A Penang Indie Fashion Brand
Threadline KL is a Penang-based indie fashion brand (anonymised name for client privacy) selling women's casual wear on Shopee Malaysia, Lazada Malaysia, and their own Shopify store. Founded in 2022, the brand built its first customer base through Penang weekend markets and Instagram, then expanded to online channels in 2024.
The numbers before AI rollout (December 2025):
| Channel | Monthly revenue |
|---|---|
| Shopee | RM18,000 |
| Lazada | RM8,000 |
| Shopify store | RM10,000 |
| Instagram DM direct | RM2,000 |
| Total monthly | RM38,000 |
The team:
- 1 owner-operator (sourcing, photography, customer service, marketing, ops)
- 1 part-time content assistant (15 hours/week — Shopee listings, social posts)
The owner was hitting a wall — couldn't ship more SKUs without hiring, couldn't run more ads without losing money, couldn't respond to WhatsApp fast enough. Classic "founder is the bottleneck" stage.
The Problem: Three Bottlenecks Stopping Growth
The three specific bottlenecks were slow listing production, slow WhatsApp response, and unprofitable Facebook ads.
Bottleneck 1 — Listings. Each new SKU took 90 minutes to list across Shopee, Lazada, and Shopify. Brand had 200 SKUs in stock but only 80 listed online. Inventory rotting.
Bottleneck 2 — WhatsApp response. Average response time was 4–6 hours during work hours, often overnight delays for after-hours messages. Customers who didn't get instant replies bought from competitors.
Bottleneck 3 — Facebook ads. Manual targeting on Meta Ads delivered CPM of RM22 — too expensive for the brand's RM65 average order value. ROAS hovering at 1.4x. Not profitable.
The AI Stack We Built
Over 14 working days, ONSET deployed a 5-tool AI stack costing RM480/month total.
| Tool | Function | Monthly cost (MYR) |
|---|---|---|
| Claude Pro | Shopee + Lazada + Shopify listing generation, WhatsApp replies | RM95 |
| ChatGPT Plus | Social captions, ad copy variations | RM95 |
| Canva Pro Team | Product visuals, ad creatives | RM45 |
| Make.com | Spreadsheet → Claude → Listings automation | RM45 |
| WATI Pro | WhatsApp Business AI chatbot | RM200 |
| Klaviyo Starter | Abandoned cart emails (already had) | RM200 |
| Total | RM480 + RM200 Klaviyo = RM680 |
Plus the existing Meta Ads spend (~RM6,000/month) migrated from manual targeting to Meta Advantage+ (Meta's native AI — no extra subscription cost).
The Three Changes That Drove Results
Change 1 — AI Listing Generation (Shopee, Lazada, Shopify)
Built a Make.com automation: owner adds a new product row to a Google Sheet (product name, features, price, photos) → automation triggers Claude to generate platform-specific listings → output saved to "ready to publish" sheet → owner reviews and bulk-uploads.
Result: time per new SKU dropped from 90 minutes to 8 minutes. Owner went from listing 10 SKUs/week to 30+ SKUs/week. Inventory backlog cleared in 7 weeks. Total active SKUs grew from 80 to 280.
Change 2 — WhatsApp AI Customer Service
WATI Pro deployed with FAQ training: 50+ pre-built responses for common questions (sizing, shipping, returns, payment, COD), photo-based size advisor ("send a photo, I'll suggest size"), order status auto-fetch from Shopify.
Result: average response time dropped from 4–6 hours to under 2 minutes. AI handles 72% of conversations end-to-end. Owner focuses on the complex 28% (custom orders, complaints, VIP customers).
Change 3 — Meta Advantage+ Ads + Klaviyo AI Recovery
Migrated all Facebook campaigns to Meta Advantage+ (Meta's AI for targeting, creative, and budget allocation). Connected Klaviyo's AI segmentation for abandoned cart sequences.
Result: Facebook CPM dropped from RM22 to RM14 (-36%). ROAS climbed from 1.4x to 2.8x. Abandoned cart recovery improved from 2% to 9% via personalised Klaviyo sequences.
The Results After 6 Months
The numbers after 6 months speak for themselves.
[IMAGE 2 — Before/after chart: 6-month results]
| Metric | Before (Dec 2025) | After (Jun 2026) | Change |
|---|---|---|---|
| Monthly revenue | RM38,000 | RM82,000 | +116% |
| Active SKUs across platforms | 80 | 280 | +250% |
| WhatsApp response time | 4–6 hours | < 2 minutes | -95% |
| Facebook ad CPM | RM22 | RM14 | -36% |
| ROAS on Meta ads | 1.4x | 2.8x | +100% |
| Abandoned cart recovery rate | 2% | 9% | +350% |
| Total monthly AI cost | RM0 | RM680 | New |
| Net margin gain | — | +RM38,000/month | — |
The team size: still 1 owner + 1 part-time assistant. No new hires. AI absorbed the volume increase.
What Made It Work (and What Didn't)
Three things that worked:
- Starting with listing automation — fastest visible win. Owner saw 200 new listings live in 8 weeks. Believed in the system.
- Migrating to Meta Advantage+ before scaling spend — fixed the unit economics before pouring more budget.
- Weekly measurement — tracked SKUs added, WhatsApp response time, ROAS, abandoned cart rate every Friday. Made wins visible.
Two things that didn't work initially:
- AI replies in pure English — Penang customers message in Hokkien-flavoured Manglish. We retrained WATI with 50 sample conversations matching local tone. Took 3 weeks to nail.
- Auto-publishing listings without review — first batch had 4 errors (wrong prices, mismatched photos). Rolled back to "AI drafts, owner reviews, then publish" workflow.
How to Replicate This Result
To replicate this Malaysian e-commerce AI result in your own brand, follow a 5-step playbook: fix unit economics first (ad ROAS), automate listings second, deploy WhatsApp AI third, recover abandoned carts fourth, measure weekly.
- Migrate Facebook ads to Meta Advantage+ for 30 days. Don't scale spend until CPM drops 20%+ and ROAS hits 2x+.
- Build the Sheet → Claude → Listings automation. First SKU takes 2 hours to set up. Every subsequent SKU takes 8 minutes.
- Deploy WATI with 50+ trained FAQ responses. Soft launch with 10% of traffic for week 1, expand from there.
- Set up Klaviyo abandoned cart sequence with AI personalisation. Recovery rate should climb to 6–10% within 30 days.
- Track 5 metrics weekly: SKUs added, response time, ROAS, recovery rate, revenue. Adjust where data shows gaps.
Typical Malaysian e-commerce SME results following this playbook: 60–120% monthly revenue growth in 6 months at AI software cost of RM450–RM700/month.
Frequently Asked Questions
How much did the AI stack cost in this case study?
The total AI stack cost RM680/month — Claude Pro (RM95), ChatGPT Plus (RM95), Canva Pro Team (RM45), Make.com (RM45), WATI Pro (RM200), and Klaviyo Starter (RM200). The brand's monthly revenue grew RM44,000 in the same period — ROI of roughly 65x on AI spend.
How long did the AI rollout take?
The full AI rollout took 14 working days from kickoff to all systems running. Measurable results appeared in week 2. Major revenue gains showed in months 2–3.
Did anyone lose their job?
No. The single part-time content assistant stayed and shifted from manual listing creation to AI review and quality control. No internal staff lost employment.
Can other Malaysian e-commerce brands replicate this?
Yes. The 5-step playbook works for any Malaysian e-commerce SME selling on Shopee, Lazada, or Shopify with monthly revenue between RM10,000 and RM80,000. Typical 6-month results: 60–120% revenue growth at the same team size.
Is this case study real?
This case study uses an anonymised composite based on common Malaysian e-commerce AI rollouts. The specific brand name ("Threadline KL") is fictional. The pattern, tools, and numbers reflect typical Malaysian e-commerce AI results we've delivered.
Want similar results for your Malaysian e-commerce business?
ONSET, the AI division of Marketing Lancers, deployed the AI stack for this brand and similar Malaysian e-commerce sellers across Penang, Klang Valley, and Johor. As an MDEC-approved provider, our work qualifies for the 50% SME Digitalisation Grant.
- Free 30-minute e-commerce AI case study walkthrough — we'll show you exactly how this maps to your business. Book a call →
- Free e-commerce AI audit — send your current Shopee/Lazada/Shopify store, get a 5-page report. Request your audit →
For the bigger picture, read our complete AI for e-commerce Malaysia guide.
Last updated: May 2026 · Written by the Marketing Lancers team
