From Ticket Queues to Self-Service for a European E-Commerce Group

A multi-brand European e-commerce group with 180+ developers across 4 brands was bottlenecked by a centralized ops team handling all infrastructure requests. We implemented a Backstage-based platform with brand-aware templates, cost attribution, and self-service workflows that reduced provisioning lead time from 5 days to 15 minutes and freed the ops team to focus on reliability instead of ticket queues.

The Challenge

The group operated 4 distinct e-commerce brands on shared infrastructure, managed by a centralized operations team of 6. Every infrastructure request — from spinning up a new environment to creating a cache cluster — went through a Jira ticket queue with a 5-day average lead time. Developers regularly worked around the queue by manually provisioning resources, creating shadow infrastructure that was untracked and ungoverned. Cloud costs were growing 40% year-over-year with no attribution to specific brands or teams. The ops team was burned out, spending 80% of their time on repetitive provisioning instead of platform reliability work. Each brand had evolved its own deployment patterns, making cross-brand knowledge sharing impossible.

Our Approach

Platform Vision & Operating Model — Weeks 1–3

Defined the target platform operating model, mapped provisioning workflows, and designed a multi-tenant Backstage architecture supporting brand isolation with shared foundations.

• Ops team interviews and ticket queue analysis (600+ tickets reviewed)
• Multi-tenant platform architecture with brand-level isolation
• Cost attribution model by brand, team, and environment
• Platform team role transition plan from ticket-takers to platform builders

Backstage Deployment & Multi-Brand Catalog — Weeks 4–6

Deployed Backstage with multi-tenant RBAC, integrated with GitLab and AWS, and built the service catalog spanning all 4 brands with ownership and cost metadata.

• Multi-tenant RBAC with brand-level permission boundaries
• GitLab integration for all 4 brand repositories and CI pipelines
• AWS resource discovery and tagging for cost attribution
• Service catalog with 150+ components across brands

Self-Service Templates & Cost Controls — Weeks 7–9

Built 18 brand-aware templates covering all common provisioning requests. Integrated AWS cost explorer for real-time budget visibility per brand.

• 18 templates covering environments, caches, queues, databases, and CDN
• Brand-specific defaults with shared security and compliance baselines
• AWS cost attribution dashboards per brand, team, and environment
• Budget alerts and approval workflows for high-cost resources

Adoption, Training & Ops Transition — Week 10

Ran adoption sprints with each brand, transitioned the ops team from ticket processing to platform engineering, and established platform health metrics.

• Brand-specific training sessions and documentation
• Ops team transition plan to platform engineering roles
• Platform health dashboard with adoption and satisfaction metrics
• Quarterly platform roadmap process driven by developer feedback

Outcomes

Provisioning Speed

Infrastructure provisioning lead time dropped from 5 days to 15 minutes. Self-service templates handled 94% of all provisioning requests without ops team involvement.

Cost Visibility & Savings

Cloud cost growth reduced from 40% to 12% year-over-year. Brand-level attribution identified $180k in orphaned resources within the first month.

Ops Team Transformation

The ops team shifted from 80% ticket work to 80% platform improvement. Two team members upskilled into platform engineering roles.

Cross-Brand Standards

All 4 brands now share deployment patterns, security baselines, and observability standards. Knowledge transfer between brands became possible for the first time.