Standard multi-vendor e-commerce stacks — built around turnkey Shopify configurations or bloated JavaScript single-page templates — collapse under real-world operating conditions across provincial West Africa. In hubs like Kenema and regional trading corridors, end users predominantly access digital stores via budget Android devices equipped with 2GB to 3GB RAM over intermittent 3G cell connections.

When an unoptimized storefront forces a customer to download 6MB of uncompressed scripts and hero banners simply to view a motorcycle listing, bounce rates exceed 70%. Metered mobile data is a tangible cost; platform success requires treating every byte as an engineering debt.

Vendors routinely snap inventory photos directly inside sun-drenched street markets using varying phone lenses, yielding 4MB to 12MB raw JPEG payloads per product. Transmitting four raw photos over spotty connections often times out the upload gateway entirely.

Lima intercepts user-selected image files directly on the client side before any network transmission occurs. Using the browser's hardware-accelerated createImageBitmap and off-screen canvas scaling, images are constrained to a maximum bounding box and encoded to next-gen WebP blobs.

async function compressVendorImage(file, maxDimension = 1200, quality = 0.75) {
  // Decode off the main UI thread
  const bitmap = await createImageBitmap(file);

  let { width, height } = bitmap;
  if (width > maxDimension || height > maxDimension) {
    const ratio = Math.min(maxDimension / width, maxDimension / height);
    width = Math.round(width * ratio);
    height = Math.round(height * ratio);
  }

  const canvas = document.createElement('canvas');
  canvas.width = width;
  canvas.height = height;
  const ctx = canvas.getContext('2d');
  ctx.drawImage(bitmap, 0, 0, width, height);

  return new Promise((resolve) => {
    canvas.toBlob((blob) => resolve(blob), 'image/webp', quality);
  });
}

One foundational lesson learned in Sierra Leonean trade environments is that commerce is conversational. Buyers rarely convert through detached multi-step credit card forms; they demand direct verification, physical inspection details, and rapid price negotiation with human proprietors.

“Rather than fighting user behavioral norms, Lima implements a dual-path checkout — automated mobile money for instant purchases, and a one-tap WhatsApp handoff for everything that still needs a human conversation.”

While automated mobile money checkout (Orange Money & Africell Afrimoney) powers instant purchases, every listing automatically synthesizes an encrypted WhatsApp direct link embedding product SKU, agreed unit price, and a uniquely traceable escrow reference.

Designing software for resilience requires deliberate technical humility. Our team derived three non-negotiable architectural axioms throughout this build:

• Aggressive Local Caching: Cache recent searches and bookmarked product cards in IndexedDB so users can review specs during cellular blackouts.
• <150KB Initial Payload Target: Strict tree-shaking, system font fallbacks, and zero unessential CSS runtime libraries.
• Geospatial Radial Clustering: Rank local merchant stock higher to eliminate long transit friction across district checkpoints.