Ecommerce Strategy & Architecture

The D2C model has become the dominant ecommerce pattern for brands formerly selling only through wholesale. It involves selling manufacturer-owned products directly to end consumers through your own storefront, cutting out intermediary retailers. The strategic advantage lies in controlling the customer relationship, first-party data collection, and brand experience end-to-end. Key operational considerations include manufacturing lead times and demand forecasting, DTC-specific fulfilment (single-item parcels vs carton packing), subscription and replenishment models, and the economic unit economics — contribution margin per order must cover ad spend (CAC), shipping, handling, and platform fees. Brands scaling D2C above $10M GMV typically need a dedicated DTC operations team, a 3PL with real-time inventory sync, and a personalisation engine that adapts to browsing patterns.

Omnichannel means selling the same product catalog across multiple channels — your own storefront, Amazon, eBay, Walmart Marketplace, Meta Shops, TikTok Shop, Instagram Shopping, Google Shopping, and wholesale portals — with a unified inventory, pricing, and order management layer. The complexity grows geometrically with each channel added: each marketplace has different listing formats, fee structures, return policies, and ranking algorithms. The operational challenge is inventory allocation — when 200 units are in warehouse and orders arrive simultaneously across 5 channels, you need real-time reservation logic to prevent overselling. A unified Order Management System (OMS) with distributed inventory, channel-aware pricing rules, and centralised returns processing is the foundational architecture. Brands above $25M GMV across 3+ channels typically need a purpose-built OMS like Manhattan, OrderCloud, or a custom headless order service.

B2B ecommerce differs fundamentally from B2C: order volumes are larger, pricing is customer-specific (account-based tiers, volume breaks, contract pricing), checkout involves purchase orders and net-terms approval, and the catalog is often segmented by customer type. A successful B2B digital transformation replaces phone-and-email ordering with self-service portals that show negotiated pricing, real-time stock levels, order history, reorder shortcuts, invoice management, and requisition workflows with multi-level approval chains. The technical architecture must support complex pricing rules engines, customer-specific catalog visibility (product X visible to Account A at $10, hidden from Account B), flexible payment methods (PO, credit card, ACH, terms), and ERP-integrated order flows. Magento and BigCommerce have strong native B2B, while Shopify requires B2B on Shopify Plus. Custom headless builds are common for $50M+ B2B organisations with SAP/Oracle ERPs.

Subscription commerce — where customers receive products on a recurring schedule — is a distinct business model with fundamentally different unit economics. The growth metric is not conversion rate but retention rate: the lifetime value of a subscriber who stays for 18 months at $40/month far exceeds a one-time purchaser. The operational complexity includes dunning logic (failed payment retry strategies), skip-a-delivery and pause features, box customisation, prepaid vs pay-as-you-go models, and upgrade/downgrade workflows. Technical challenges involve recurring payment tokenisation with Stripe Billing or Braintree, subscription lifecycle management, and inventory forecasting that must account for renewal volume. Brands scaling subscriptions above 10K active subscribers need a dedicated subscription platform (Recharge, Ordergroove) or a custom subscription engine integrated with the core commerce stack.

Order Management Systems (OMS) are the operational backbone of any ecommerce business above $10M GMV. When an order is placed, the OMS must reserve inventory across multiple warehouses in real-time, route the order to the optimal fulfilment location based on shipping cost and proximity, generate pick lists and shipping labels, trigger 3PL integration (ShipStation, Fulfillment by Amazon, or custom WMS), update tracking across all customer touchpoints, and handle exceptions (partial fulfilment, backorders, address validation failures). The inventory orchestration layer must handle distributed stock across physical stores, warehouse locations, and 3PL partners — implementing reservation logic that prevents overselling while maximising sell-through. Event-driven architecture (Kafka/Kinesis) is essential here: the OMS subscribes to order-placed events and publishes fulfilment events to downstream systems, ensuring the entire pipeline is asynchronous, resilient, and observable.

Fulfilment automation spans rate shopping (comparing carrier rates across UPS, FedEx, USPS, DHL, and regional carriers), address validation and auto-correction (ShipEngine, Loqate), shipping label generation, customs documentation for cross-border orders, and tracking webhooks that update customers proactively. Advanced automation includes split-shipment logic (when one item is in warehouse A and another in warehouse B, the system optimises either splitting the shipment or transferring inventory based on cost), dropship vendor routing (orders routed directly to manufacturers for direct-to-customer shipping without holding inventory), and buy-online-pickup-in-store (BOPIS) workflows that integrate with retail POS systems. At Cosmo Music, I implemented an omnichannel inventory system synchronising real-time stock across 45+ physical stores and the online platform, reducing overselling incidents by 95% and enabling ship-from-store.

Marketing automation in ecommerce is event-driven lifecycle marketing — not generic email blasts. The customer journey is mapped to behavioural triggers: browse abandonment (user viewed product X but did not add to cart), cart abandonment (added to cart but did not purchase within 2 hours), post-purchase sequences (order confirmation, shipping notification, delivery confirmation, review request, replenishment reminder), win-back flows (no purchase in 90 days), and VIP segmentation (top 10% customers by LTV receive early access to launches). The technical integration involves webhook subscriptions from the commerce platform to the marketing automation tool (Klaviyo, Braze, HubSpot, or custom), real-time event streaming for behavioural triggers, and dynamic content blocks that personalise emails based on browse history, purchase history, and predictive next-order date.

Beyond custom-engineered automation, many ecommerce operations leverage no-code/low-code integration platforms (Zapier, Make, n8n, WorkiP) to connect their tech stack without writing custom middleware. Common automation flows include: new Shopify order triggers a row in Google Sheets and a Slack notification; low inventory threshold triggers an alert to the purchasing team; new customer review posts to Instagram Stories; and weekly sales summaries generate reports in Looker. These tools are excellent for rapid prototyping and small-volume workflows but have limitations at scale: per-task pricing can get expensive above 50K tasks per month, latency is measured in minutes not seconds, and error handling is often opaque. For production-grade automation above 5K orders per day, I recommend migrating critical flows to custom backend functions with proper monitoring, alerting, and retry logic.

Checkout is the single most critical conversion point in ecommerce — every friction point in the checkout flow directly reduces revenue. The optimisation discipline includes guest checkout (never force account creation), accelerated payment methods (Apple Pay, Google Pay, Shop Pay, PayPal Express) that bypass manual form entry entirely, address auto-complete (Google Places API), one-page checkout for mobile, and progressive disclosure (showing only the fields needed for the current step). Payment processing optimisation includes multi-acquirer routing (routing American Express through one processor and Visa through another for lower fees), tokenised vault storage for recurring payments, 3D Secure/SCA compliance for European cards, and local payment methods (iDEAL in Netherlands, Bancontact in Belgium, UPI in India) for cross-border conversion. Brands above $20M GMV with 3+ payment methods typically need a payment orchestration layer (Adyen, Braintree, or custom) to unify reporting and routing.

Conversion Rate Optimisation (CRO) is the systematic discipline of improving the percentage of visitors who purchase. It involves funnel analysis (identifying where users drop off between landing, product page, cart, and checkout), A/B testing (comparing two variants of a page element and declaring a winner based on statistical significance), heatmapping (Hotjar, FullStory, recording user scroll depth and click patterns), and session recording to identify UX friction points. The analytics architecture includes event tracking (every user interaction as a structured event: product_viewed, add_to_cart, begin_checkout, purchase), attribution modelling (determining which marketing channel gets credit for the conversion — last-click, multi-touch, data-driven), cohort analysis (grouping users by acquisition source and tracking their behaviour over time), and real-time dashboards (Looker, Tableau, or custom React dashboards). The customer data platform (CDP) layer — Segment, mParticle, or a custom event router — is the foundational data infrastructure that feeds all downstream systems.

Ecommerce security is a regulatory and operational requirement, not an optional consideration. PCI DSS compliance (Payment Card Industry Data Security Standard) mandates how cardholder data is handled — most platforms achieve this through SAQ-A (fully delegating card data to a payment processor via hosted fields or redirect) rather than the more burdensome SAQ-D (storing card data yourself). GDPR compliance requires consent management for cookies and tracking pixels, data subject access requests (exporting a user's data), and right-to-erasure (deleting a user's data across all systems). Fraud prevention involves velocity checks (flagging orders with mismatched billing/shipping addresses, high-frequency purchases from one IP, or unusual order values), AVS/CVV checks on payment processing, and ML-based fraud scoring (Sift, Signifyd, or custom models). At Lorex Technology, I implemented data privacy and encryption standards for GDPR and HIPAA compliance across a $200M+ revenue platform.

Page speed is a direct revenue lever in ecommerce — Google research shows that improving page load time from 3 seconds to 1 second increases mobile conversion by 27%. Core Web Vitals (Largest Contentful Paint, First Input Delay, Cumulative Layout Shift) are the Google-defined metrics that determine search ranking and user experience. Performance optimisation techniques include image optimisation (WebP/AVIF formats, responsive srcset, lazy loading for below-the-fold images), code splitting (only loading the JavaScript needed for the current route), CDN edge caching (serving static assets from the nearest POP), server-side rendering or static site generation (pre-rendering product pages at build time), and preloading critical resources (hero images, fonts, primary CSS). Technical SEO includes structured data (JSON-LD for Product, Offer, BreadcrumbList schemas), canonical URLs, sitemap management, and crawl budget optimisation.