Load Testing reveals risks following re-platform to composable commerce

Why autoscaling alone isn’t enough to protect CX at peak

A leading high street jewellery retailer re-platformed to a composable storefront with autoscaling, designed to handle seasonal peaks. Rather than assuming the setup was bulletproof, they took the smart approach, commissioning journey based Load Testing to validate real-world performance ahead of their busiest trading periods.

When autoscaling creates hidden risks

The test delivered exactly what was needed: peace of mind and actionable insight. It revealed a hidden autoscaling configuration problem that triggered spikes in error rates, including HTTP 500 errors, missing content, and empty product listings, every time the site scaled.

Once traffic reached around 3,000 concurrent users, these errors escalated and no longer recovered between scale events. Left undiscovered, this could have left customers unable to purchase products, facing empty product listings, failed product pages and failed product filters, during the busiest sales periods.

BFCM risks you won’t uncover with conventional Load Testing

Conventional Load Testing often stops at measuring server response times and error rates. While those metrics are valuable, they wouldn’t have revealed the real-world customer experience issues uncovered here, problems that directly impact conversion but don’t always trigger widespread errors.

Real-world journey based Load Testing exposed subtle but critical issues such as:

• Empty product listing pages (PLPs) where the grid loaded but no products appeared

• Failed product filters that stayed stuck in a loading state

• Missing third-party content that never populated

Test and Retest

Because the problem was found early, the fix was made with no customer impact. A retest proved the change worked: the site scaled cleanly to all nodes and handled traffic well beyond peak levels, maintaining stable journey delivery times without sustained errors or degradation over time. The consistency of performance under heavy load was a marked improvement over their previous SAP Hybris platform.

The takeaway for any retailer preparing for BFCM?

Autoscaling is a valuable safety net but it’s no substitute for proactive, real-world Load Testing to uncover CX issues, protect customer journeys, and ensure your busiest trading days run without a hitch.

---

Composable Commerce: One Concept, Two Playbooks

“Composable commerce” is an approach where retailers build their eCommerce stack from best-in-class components, rather than relying on a single monolithic platform.

Gartner’s approach: Emphasises packaged business capabilities (PBCs) — self-contained features (like search, payments, CMS) that can be swapped in or out as needs change.

MACH Alliance’s approach: Advocates for systems that are Microservices-based, API-first, Cloud-native, and Headless. The focus is on openness, flexibility, and vendor neutrality

In practice, many retailers mix and match elements from both schools of thought, selecting components they like best and integrating them into a unified storefront

---

Further resources

Planning for BFCM?

Your peak season will be here before you know it. Load Testing now means you can uncover and fix hidden performance risks while there’s still time to act. Whether it’s validating autoscaling, stress testing a new platform, or proving your infrastructure can handle peak traffic, our real-world journey Load Testing gives you the data and confidence you need to deliver flawless CX when it matters most.

Best practices for AWS autoscaling

Learn more about best practices for AWS Autoscaling.