The Last Mile Is Now a Software Problem

Checkout used to be the finish line.

It is not anymore.

The moment a customer completes a purchase is now the beginning of the most consequential part of their experience. Delivery accuracy, communication quality, return ease, and post-purchase support are where loyalty is built or quietly lost. And increasingly, they are where the technical debt of underinvested fulfillment systems becomes visible, not in dashboards, but in churn.

Roughly 70% of shopping carts are abandoned before checkout. The attention of most commerce teams goes there, understandably. 

But the remaining 30% that complete a purchase represents a moment of genuine trust. What happens next determines whether that customer ever comes back.

The brands winning long-term retention have stopped treating fulfillment as an operational problem. They treat it as a software problem. And the difference in outcomes is measurable.

Why fulfillment became an engineering discipline

For most of e-commerce history, the post-purchase experience was handled by logistics providers, warehouse management systems, and operations teams. The software layer was thin: an order management system, a handful of carrier integrations, and email notifications.

That model worked when customer expectations were calibrated to two-day shipping and weekly tracking updates.

Those expectations no longer exist.

Same-day and next-day delivery have shifted from premium features to competitive baseline requirements across most major categories. Real-time order tracking is now expected, not appreciated. Returns need to be seamless and fast or they drive permanent churn. And customers who experience a fulfillment failure once are significantly less likely to return, regardless of how good the product experience was.

The implication for engineering teams is significant. Fulfillment is no longer a downstream concern that software touches lightly. It is a product surface that requires the same architectural rigour as checkout, search, and personalization.

The Mercadão case: building for demand that multiplies

Mercadão is Portugal's first two-hour delivery marketplace, connecting customers with a range of retailers for on-demand delivery across multiple categories. The promise is made fast. The engineering requirement is faster.

The challenge. Mercadão needed to build a platform capable of coordinating high-volume, time-sensitive deliveries across multiple retailer partners and courier networks, all in real time. Inventory had to be accurate at the moment of purchase. Order routing had to be dynamic, adapting to courier availability, proximity, and load. Partner integration had to be flexible enough to onboard new retailers without rebuilding core systems. And the entire platform had to absorb significant demand growth without degrading the delivery promise.

The decision. Glazed built the platform on a stack designed for real-time operations: Angular and React on the front end, AWS infrastructure, Kubernetes for orchestration, and Node.js for the services layer. The architecture separated the order management logic from the inventory layer and the routing engine, allowing each to be scaled and updated independently. A dynamic routing system was built to allocate deliveries based on live courier data rather than static rules. Real-time inventory sync was implemented across partner retailers, ensuring that availability shown to customers reflected actual stock. A partner API layer was built to enable new retailers to integrate with the platform without requiring custom engineering on each side.

The impact. A platform that could grow with demand without requiring architectural rebuilds at each growth stage. When volume surges, the system scales horizontally. When a new retailer joins, the integration is handled through the partner API layer. When routing conditions change, the engine adapts in real time rather than relying on manual intervention.

The Mercadão platform is a working example of what fulfillment engineering looks like when it is treated as a product investment from the beginning rather than a problem to be solved after scale creates pressure.

The four systems that define a resilient fulfillment platform

Across the commerce work Glazed has done in delivery, marketplace, and on-demand retail; four systems consistently determine whether a fulfillment platform performs at scale or breaks under it.

Real-time inventory orchestration. Inventory accuracy is the foundation of the entire post-purchase experience. A customer who purchases an item that is not actually available experiences the worst possible outcome: a failed promise at the moment of highest trust. Platforms that rely on batch inventory updates, rather than real-time synchronization, carry structural accuracy risk that compounds with catalog size and order volume.

Dynamic order routing. Static routing rules degrade as operational conditions change. Courier availability fluctuates. Partner fulfillment capacity varies by time and day. Geographic demand patterns shift. A routing engine that adapts to live operational data rather than fixed rules is significantly more reliable and more efficient than one that requires manual adjustment.

Post-purchase communication infrastructure. The most common fulfillment failure is not a delivery that goes wrong. It is a delivery that goes right but leaves the customer uncertain. Proactive, accurate, and timely communication at each stage of the fulfillment journey reduces customer support volume, increases satisfaction, and builds the trust that drives repeat purchases.

Returns as a first-class product surface. Returns are the most emotionally loaded moment in the post-purchase experience. A friction-free return process converts a potentially negative experience into a demonstration of reliability. Platforms that engineer returns as a core product capability, not as an edge case handled by customer support, consistently outperform those that do not on retention metrics.

The strategic question commerce teams are not asking often enough

Most commerce investment conversations focus on acquisition: traffic, conversion, paid media, and SEO. The post-purchase experience receives a fraction of that attention and investment.

The data does not support that allocation.

Acquiring a new customer costs five to seven times more than retaining an existing one. Returning customers spend, on average, 67% more than first-time buyers. And the single most reliable predictor of whether a customer returns is whether their first fulfillment experience met or exceeded the promise made at checkout.

The engineering question, then, is not whether to invest in fulfillment infrastructure. It is whether the investment you are making today is proportional to the retention upside it enables.

Before your next roadmap cycle, three questions are worth asking with precision:

1. What percentage of your post-purchase support volume is driven by fulfillment failures or communication gaps that better engineering could prevent?
2. How long does it take your platform to reflect an inventory change across all customer-facing surfaces?
3. If your delivery volume doubled tomorrow, which part of your fulfillment stack would break first?

The answers define where the engineering investment needs to go.

Closing the series

This series has mapped three layers of the same challenge: building commerce platforms that grow without breaking.

The structural foundation is architecture. The operational constraint is performance. The retention lever is fulfillment.

These are not three separate problems. There are three dimensions of a single engineering discipline: building commerce infrastructure that is worthy of the customer trust it is asking for.

The companies that win the next decade of commerce will not necessarily be the ones with the most innovative products or the largest marketing budgets. They will be the ones whose engineering teams understood, earlier than their competitors, that commerce is now a software discipline and built accordingly.

Series recap:

• Article 1: E-commerce in 2026: The End of "Just a Store". The structural shift from storefront to distributed platform and the architecture that enables it.
• Article 2: Scaling E-commerce Systems Without Breaking the Experience. The engineering decisions that preserve performance as traffic, catalog, and transaction volume grow.
• Article 3 (this one): Why fulfillment is the new product surface and what resilient last-mile infrastructure looks like in practice.

Glazed builds and scales commerce platforms for companies where engineering quality directly determines business outcomes, including Farfetch, Gopuff, and Mercadão. If the post-purchase layer of your platform is the part of your roadmap that keeps getting deprioritized, let's have a direct conversation about what it would take to change that: info@glazedsolutions.com.

Thanks for reading. If you enjoyed our content, you can stay up to date by following us on X, Facebook, and LinkedIn 👋.