Gamified microlearning that turns store salespeople into certified brand specialists
LXP built with Next.js 15 and React 19 for Marken Fassi — 10 to 60 second microlearning consumable between customers, gamification with streaks and tiers from Bronze to Black Ruby, FASSI.AI sales assistant with a structured answer base from the catalog, private community for sharing displays among retailers, team performance dashboards and a certification program that positions salespeople as official brand specialists. All in a mobile-first interface designed for the shop floor, with fast loading even on unstable 3G/4G connections.
- 28%
- Sales Increase (est.)
- 67%
- Daily Engagement (target)
- 78%
- Certification (pilot)
Client
Marken Fassi
Sector
Premium retail · Bedding
Timeline
3 meses · 2024-03
Live application
Open applicationThe challenge
Friday, 2 p.m. Ana, a sales associate at a multi-brand store in São Paulo, is helping a customer who asks for the softest bedding set for winter. For a moment she hesitates — there are dozens of products, yarns, weights and blends. Instead of calling the supervisor, she opens CASA FASSI. In 45 seconds, a micro-module reminds her of the fabric differences, suggests two options and links to a quiz she can finish before the customer tries the first piece. When the sale closes, the ticket tops R$ 1,800. Ana earns points, levels up and, next week, will be listed as a certified specialist.
O varejo premium de enxovais enfrentava vendedores de lojas multimarcas com conhecimento técnico limitado sobre tecidos, fios e diferenciais da marca, resultando em atendimento genérico e baixo ticket médio. Havia pouca retenção de conhecimento após treinamentos pontuais, ausência de conexão entre vendedores e a marca, dificuldade em escalar o relacionamento com centenas de lojistas espalhados pelo Brasil e falta de reconhecimento contínuo para a excelência em vendas.
The approach
We treated learning as a product, not content. First we mapped the salesperson's real journey: idle moments during shifts, quick customer interactions and decision points that define the ticket. From there we designed a microlearning architecture in 10 to 60 second bursts, punctuated by gamification, daily challenges and immediate feedback. The reward system was designed to compete with wellness apps: streaks, tiers, visual achievements and real benefits. AI enters as a coach at the exact moment the seller feels doubt, not as a replacement for the trainer.
The solution
CASA FASSI is a private Learning Experience Platform (LXP) for sales associates and retailers. It connects training, community, performance and recognition into an ecosystem where learning and selling happen in the same flow.
- Microlearning in 10 to 60 second bursts, consumable between customers, with structured tracks by fabrics, collections and sales techniques — each module designed to fit the real idle window of the salesperson on the shop floor
- Gamification with daily streaks, tiers from Bronze to Black Ruby, competitive leagues, achievement badges and real benefits like early access to collections and public peer recognition
- FASSI.AI: sales assistant with a structured answer base from the catalog, brand script and sales context, designed to guide the seller's reasoning at the exact moment of doubt — with a roadmap toward LLM and RAG
- Private community for sharing displays, sales cases, technical questions and best practices among salespeople from different stores, creating a peer-to-peer learning ecosystem
- Performance dashboards so retailers can track teams, progress per track, certification rate, internal ranking and weekly engagement — with aggregated views by store and by seller
- Certification program that positions salespeople as official brand specialists, with validated badges per completed track and proficiency levels recognized by Marken Fassi
Timeline
Como o projeto foi conduzido, do mapeamento ao deploy. How the project was conducted, from mapping to deploy.
Semana 1-2
Mapeamento
- Immersion on the shop floor with sellers and retailers
- Mapping the real journey between customer interactions
- Identifying idle peaks and moments of doubt
- Defining baseline metrics for retention and engagement
Semana 3-10
Desenvolvimento
- LXP architecture with 10 to 60 second microlearning
- Gamification implementation with tiers and streaks
- FASSI.AI development with structured answer base
- Weekly validation with real seller cohorts
Semana 11-12
Lançamento
- Controlled launch with pilot retailers
- Fine-tuning rewards and microcopy
- Real-time KPI monitoring
- Onboarding the first wave of certified sellers
Methodology
Como o projeto foi conduzido, do mapeamento ao deploy. How the project was conducted, from mapping to deploy.
Map the seller's real journey between customer interactions
Validate assumptions with retailers and floor salespeople
Document decisions in a shared decision log
Iterate weekly using retention and engagement metrics
Run A/B tests on microcopy, timing and rewards
Guided onboarding with pilot retailers before general launch
Architecture
- → SSR container with Next.js 15 to deliver pages and cache
- → Serverless API for authentication, gamification and analytics
- → PostgreSQL container for users, retailers and progress data
- → Redis container for session cache, leaderboards and queues
- → OpenAI container to orchestrate RAG and response generation
- → Workers container for notification pipelines and async jobs
Conceptual diagram — does not reflect real implementation
Tech stack
Technical choices and what was left out.
Frontend
Why: SSR, streaming and App Router to deliver lessons in milliseconds
Rejected alternative: Astro — strong for static content, but not for rich interactivity
Why: New hooks and concurrency for gamification animations
Rejected alternative: Vue 3 — mature, but less Server Components ecosystem
Why: Fast visual standardization and a small CSS payload
Rejected alternative: Styled Components — adds runtime and increases bundle size
Backend
Why: Serverless in the same repo, reducing latency and operations
Rejected alternative: Express on a VM — requires more infrastructure and has higher latency
Why: Fully typed APIs, reducing errors between frontend and backend
Rejected alternative: Hand-written REST — more boilerplate and less type safety
Why: Lightweight state management with local persistence for offline progress
Rejected alternative: Redux Toolkit — more boilerplate and a steeper learning curve
Data/Cache
Why: Consistent relational data for users, retailers and progress
Rejected alternative: MongoDB — flexible, but less predictable for joins and transactions
Why: Session cache, leaderboards and queues with fast data structures
Rejected alternative: Memcached — limited to simple key-value storage
Why: Typed ORM and reliable migrations
Rejected alternative: Sequelize — less type safety and worse DX
AI/Infra
Why: Engine behind FASSI.AI for RAG and brand-toned response generation
Rejected alternative: Self-hosted open-source LLM — unacceptable infrastructure cost and latency
Why: Low cold start and geo-distributed execution
Rejected alternative: AWS Lambda — higher cold starts and heavier configuration
Why: Workers for async pipelines and notifications
Rejected alternative: Long-running servers — high cost and waste for sporadic jobs
Decision log
What I chose, what I rejected, and why.
Microlearning instead of long courses
Why: The multi-brand salesperson has only a few idle minutes per day. 30-minute courses caused drop-off. 10 to 60 second fragments fit the real floor routine and double return frequency.
Rejected alternatives: We evaluated a traditional LMS and recorded videos, but historical adoption was low and knowledge retention was not measurable.
AI as coach, not an answer bot
Why: FASSI.AI was designed to guide reasoning: it asks, gives technical context and lets the salesperson answer. The current version uses a structured answer base from the catalog, with a roadmap toward LLM/RAG.
Rejected alternatives: A simple FAQ chatbot would give fast answers, but would not train the salesperson to argue and close with confidence.
Gamification with tangible benefits
Why: Points and leaderboards alone do not sustain long-term engagement. We linked tiers to prizes, public recognition and early access to collections, turning learning into career and status.
Rejected alternatives: We tested pure scoring and seasonal challenges, but without real rewards the retention curve dropped after the second month.
FASSI.AI
FASSI.AI — assistant design (planned RAG evolution)
Seller asks a question about a product or sales scenario
System identifies context: catalog, journey and seller history
Current version: structured answer base from catalog and brand script
Roadmap: RAG engine with LLM for generative responses in brand tone
Validation and guardrails before reaching the seller
Respostas validadas contra catálogo. Fallback para busca estática. Cache de 24h. Responses validated against catalog. Fallback to static search. 24h cache.
Gamification
Sistema de motivação estruturado para vendedores. Structured motivation system for sales teams.
XP
Leagues
Badges
Streaks
Leaderboards
Missions
Coins
Certifications
Gamification
Security and access
Security and access governance
-
Distinct permissions for seller, retailer and admin roles
-
SSO for centralized login
-
Content moderation and reporting in the community
-
Data segmentation: each retailer only sees their own team
-
Secure integrations with catalog and gateways
-
LGPD compliance for handling seller data
Results
Client-reported results during rollout
Baseline: 22%
Fonte: Marken Fassi sales report — Q2 2024
Baseline: 41%
Fonte: LXP analytics — 30-day average
Baseline: 18%
Fonte: Retailer retention — 6 months post-launch
Baseline: R$ 1,440
Fonte: Transaction data — pre/post comparison
Baseline: 31%
Fonte: CASA FASSI certification platform
Baseline: 8s
Fonte: Assistant latency metrics — 7-day average
Client-reported data
Lessons learned
Not everything went as planned.
— The biggest lesson was that retention does not depend on more content, but on more context. Microlearning works when it is triggered at the exact moment of doubt. Gamification works when there is real status. And AI works when it amplifies the salesperson, not when it tries to replace them. I also learned that validating with real retailers in the second week saved months of adjustments.
Ready to build something similar?
I transformed a complex ecosystem into an intuitive platform that scales with the business. I can do the same for yours.