Business context
Moulpochetat is an impulse-buy / mystery-pack e-commerce funnel operating cash-on-delivery (COD) in Morocco and the broader MENA region. The core product: a curated mystery box containing a refurbished iPhone (models 11 through 16, Pro and Pro Max) at a price point significantly below retail. COD pricing is fixed at 890 MAD (~$89 USD). The market: Morocco, Tunisia, UAE, Saudi Arabia, with primary traffic from Morocco.
The business model depends on two levers: (1) traffic cost — can we get clicks and form-fills cheaply, and (2) delivery rate — what percentage of COD orders actually convert when the delivery agent calls. Delivery rate averages 65-71% depending on confirmation source. Traffic cost — CPA per confirmed lead — is what this case study focuses on.
Meta Ads is the primary acquisition channel, representing 85% of monthly spend. The funnel is: Meta creative → Shopify landing page with country selector → mystery pack form (name, phone, address, model preference) → WhatsApp confirmation flow → courier dispatch.
This case study covers January through March 2026 — the transition from pre-CreaScale hand-briefed creative to full PDA workflow.
The problem before CreaScale
Pre-CreaScale, Moulpochetat ran Meta Ads using creative briefed to a freelance designer (roughly $80 per creative, 2-3 day turnaround) with copy written by the founder in French and adapted manually to Arabic / Darija. Typical cadence: 4-6 new creatives per month.
Three structural problems emerged through late 2025:
- CPA creeping: Q3 2025 CPA was $3.10 confirmed lead. By Q4 2025 it had drifted to $4.20 — a 35% increase with no corresponding increase in delivery rate. Something was breaking in the top of funnel.
- CTR below benchmark: Feed CTR sat at 1.1%, well below the 1.5%+ benchmark for MENA COD. Hook rate on video creatives was 28% — deep into "mediocre" territory.
- Creative fatigue hitting fast: Each new creative was winning for 8-12 days before CTR decayed 30%+. The 4-6 new creatives per month cadence couldn't keep up.
The diagnosis was clear: creative angles were stuck. The freelance designer was producing competent variations of the same "luxury phone, 890 MAD, mystery" concept. Different colors, different backgrounds — same angle. Meta's Andromeda delivery system was treating all variations as one optimization signal and learning slowly.
The alternative options reviewed: hire a second designer ($80 × more creatives = linear cost scaling, same angle problem), switch to AdCreative.ai ($159/month, template-variation tool = same angle problem), or try CreaScale's PDA workflow (one prompt → a winning UGC video ad + static ad variants, from the Starter plan at $19/mo — the most accessible subscription entry point, cancel anytime).
CreaScale setup — the workflow
The switch happened in two phases: pilot in January 2026, full migration February-March.
Phase 1 — January pilot
Three CreaScale runs on the Starter plan ($19/mo, 360 monthly credits, one wallet for videos + static ads). Inputs: three different product variants (iPhone 13 Pro pack, iPhone 15 pack, iPhone 16 Pro Max pack). Each run produced a UGC video ad (AI avatar talking-head + voice-over + motion) plus matching static ad variants, each mapped to a Persona × Desire × Awareness combination — spanning cold/warm personas, six desires (Status, Safety, Comfort, Time, Money, FOMO), and five awareness levels.
Every output shipped with multilingual copy: French (primary), Arabic (MSA for formal, Darija code-switching for casual), English (GCC fallback). Total: 3 UGC video ads + their static variant sets + copy in 15 minutes of founder time vs 6 weeks of designer/copywriter work.
Pilot test: 3 new Meta ad sets, $30/day each for 5 days, $450 total spend. Result: pilot CPA $2.10 — a 50% improvement vs $4.20 baseline on the same products.
Phase 2 — full migration February-March
Full workflow rollout:
- Input: Shopify product URL (e.g., /products/mystery-pack-iphone-13-pro). CreaScale scrapes the product page + analyzes Meta Ad Library competitors.
- Output: one prompt → a winning UGC video ad + static ad variants. CreaScale picks the i2v model, generates a UGC avatar talking-head with voice-over and motion, then cuts static ad variants — each mapped to a different Persona × Desire × Awareness combination, with multilingual copy.
- Meta campaign: 8 ad sets (1 per angle), Advantage+ Shopping Campaigns structure, $20-30/day per ad set, CBO at campaign level.
- Rotation: a fresh UGC video ad + static variants per product variant every 14 days. At the Pro plan ($49/mo) the recurring creative budget settled around $49-149/month across all variants — a fraction of the freelance designer cost, with each fresh UGC video ad + its static set running roughly $6-7 of plan value.
- Fatigue management: angle-level CTR tracked daily; when an angle dropped 25% from peak, it was paused and a fresh angle from the next CreaScale run replaced it.
By mid-February, the 8-ad-set structure with fresh rotations was stable. By March, CPA had landed at $1.65 consistently.
Before / after — the numbers
| Metric | Before | After (3 mo) | Delta |
|---|---|---|---|
| Meta CPA (confirmed lead) | $4.20 | $1.65 | −61% |
| CTR (feed placements) | 1.1% | 2.8% | +155% |
| Hook rate (video creatives) | 28% | 47% | +68% |
| ROAS (platform-reported) | 1.8x | 3.6x | +100% |
| Frequency (30d ad set avg) | 4.2x | 2.6x | −38% |
| Creative cost / month | $320-480 (designer) | $49-149 (CreaScale plan) | −66% |
| Time from brief to launch | 2-3 days | 15 minutes | −99% |
| Delivery rate (COD) | 66% | 69% | +3pt |
| Unique angles in rotation | 2-3 | 8-12 | +4x |
Lessons learned
- PDA angles beat template variations — The 50% CPA drop in the pilot came from testing genuinely distinct psychological stances, not from better-looking creatives. Meta's Andromeda learns faster from 8 signals than 8 reskins.
- COD-specific angles work — The "FOMO × Most-aware" angle ("Last 48 iPhone 15 packs — 890 MAD") outperformed all generic "Luxury mystery" angles by 2.5x CPA. COD funnels reward urgency angles more than brand angles.
- Multilingual is underrated leverage — Native Arabic / Darija code-switching copy vs French-translated-to-Arabic increased CTR 22% in Morocco. For MENA markets, this is 1-2 months of CPA improvement alone.
- Frequency down = headroom up — Frequency dropped from 4.2x to 2.6x as 8-angle rotation replaced 2-3 angle cycling. Lower frequency = less fatigue = more runway for the same audience.
- Founder time is the real savings — The 2-3 day designer cycle × 6 creatives/month = ~18 days of founder attention per month pre-CreaScale. Post-CreaScale, it's ~2 hours/month. That time now goes to funnel optimization and delivery confirmation — both compounding revenue levers.
"The 50% CPA drop in the first week was honestly shocking — I expected maybe 10-15% improvement and it blew past that. But what matters long-term is that my founder-time on creatives went from 18 days a month to 2 hours. I got my afternoons back. That is the actual win."— Abderrahmane Bouabdli, Founder, Moulpochetat (and CreaScale AI)
What's next
The Moulpochetat playbook is being extended in Q2 2026 to two new SKU clusters (iPhone 14 pack + dedicated "Pro Max only" pack) and two new markets (Algeria + Tunisia). Target CPA in new markets: $2.00-$2.50 — accepting a modest cold-start penalty while the funnel tunes. Next case study update (July 2026) will cover the scale-up from $12K/month to $40K/month Meta spend.
Related reading
- The PDA Framework — how CreaScale generates distinct angles
- Meta Ads Creative Strategy 2026 — pillar guide
- Best 15 AI Ad Creative Tools 2026
- Dropshipping Ad Creative Automation Workflow