Retail
Scalable Ecommerce Platform for Retail Brands, USA
Increased conversion rate by 35%, reduced cart abandonment by 28%, and enabled the platform to handle 10× more concurrent users.
Delivered in approximately two months by a small dedicated team.
Project Highlights
Scalable Architecture
Built for growth — Fixed Price designed to scale
Secure & Compliant
Enterprise-grade security and data protection standards
Measurable ROI
Informal per-store WhatsApp with no records → Integrated messaging resilient to load shedding
South Africa Delivery
Delivered for a South Africa-based Retail client
Tech Stack
Industry
Retail
Sector
Regional Grocery & General Retail Chain
Region
South Africa
Project Type
Fixed Price
This project was delivered under a Non-Disclosure Agreement. The client operates in the Retail sector (Regional Grocery & General Retail Chain) and is based in South Africa.
WhatsApp is the dominant messaging channel in South Africa, and the chain's customers were already messaging individual store numbers informally with no record kept. The business wanted to formalise that, but the constraints were unusual. Load shedding interrupts store connectivity on a scheduled but disruptive basis, so any integration assuming a store is continuously online would fail predictably and often. Customer data costs matter — a substantial part of the customer base is price-sensitive about mobile data, so message payloads had to stay light and image-heavy promotional sends were a bad idea. POPIA governs direct marketing consent, and the chain had no reliable consent record for the customers already messaging them, so the consent position had to be rebuilt properly rather than assumed. The POS system was an older on-premise product per store with a database that could be read but no meaningful API, which ruled out a clean real-time integration and required something more defensive.
Informal per-store WhatsApp with no records
Integrated messaging resilient to load shedding
We built a Node.js service handling the WhatsApp Business API with MongoDB storing conversation state and message history. The POS integration was designed around intermittent connectivity from the start: a lightweight agent at each store reads order and collection status and pushes to the central service with a durable local queue, so a store losing power or connectivity accumulates events and syncs when it returns rather than losing them. Nothing in the design assumes a store is reachable at any given moment. Consent was rebuilt from zero — customers opt in explicitly through a documented flow with the consent record stored against their number, and opt-out is honoured immediately and permanently, which is what POPIA's direct marketing provisions require rather than an inherited list. Message templates were kept text-first and deliberately light on media to respect customer data costs, with promotional sends rate-limited and segmented rather than blasted. Order collection notifications are transactional and go out as status changes arrive from the store agent. Store staff have a simple console to see and respond to customer messages for their own store only. We built monitoring around the store agents specifically, so head office can see which stores are currently offline and for how long, rather than discovering a sync failure days later through missing notifications. Related: custom software development and our ecommerce and retail work.
Delivered in approximately two months by a small dedicated team. Collection notifications now reach customers automatically as orders become ready, replacing informal per-store messaging with no record. The durable queue design has proven itself — stores routinely go offline during load shedding and no notification events have been lost, which was the specific failure the client feared most. A properly documented consent base was built from scratch, putting the chain in a defensible POPIA position for direct marketing where previously it had none. Message engagement is strong, consistent with WhatsApp's position in the South African market, and text-first templates kept payloads light enough to avoid customer data-cost concerns.
Transformation Summary
Informal per-store WhatsApp with no records → Integrated messaging resilient to load shedding
This project was delivered under a Non-Disclosure Agreement (NDA). Specific client details, business data, proprietary workflows, and project credentials remain confidential. Screenshots and live URLs are available upon mutual NDA execution.
Request an NDA to see full detailsIt would if it assumed stores were continuously reachable, which is why nothing in the design does. Each store runs a lightweight agent with a durable local queue, so a store losing power accumulates events and syncs when it returns rather than losing them — no notification events have been lost since go-live.
Let's discuss your project and create a success story together.
Retail
Increased conversion rate by 35%, reduced cart abandonment by 28%, and enabled the platform to handle 10× more concurrent users.
Education
Live across all three campuses for the start of the academic year, delivered in approximately three months by a small dedicated team.
Education
Delivered in approximately four months by a small dedicated team, live in time for the admissions intake.