How a 120-Engineer SaaS Company Cut Service Onboarding from 3 Days to 2 Hours with Backstage Golden Paths
A mid-size B2B SaaS company replaced a fragmented self-service mess with a Backstage IDP and opinionated golden paths, dropping new-service onboarding from 3 days to under 2 hours and halving platform support tickets.
Architecture · intermediate · 12 min read · 2359 words · By Sandip Das · Published
Composite scenario: built from common industry patterns; the figures are illustrative.
What this covers
- The Situation
- The Problem
- Constraints
- Options Considered
- The Solution
- Implementation
- Results
- What Went Wrong Along the Way
- Lessons Learned
- Apply It Yourself
Key numbers
| Metric | Before | After |
|---|---|---|
| New service onboarding time | ~3 working days | ~2 hours |
| Platform Slack support messages per week | 40–55 | 12–18 |
| Services behind base image by 2+ minor versions | 34% | 4% |
| Time for new engineer to find canonical docs | 4–6 hours | 20–30 minutes |
| Platform team toil (FTE equivalent) | ~1.5 FTE | ~0.4 FTE |
Key components
- Backstage Software Catalog — Single source of truth for all services, their owners, dependencies, and metadata, ingested from catalog-info.yaml files in every GitHub repo.
- Backstage Scaffolder + Golden Path Templates — Nunjucks-based cookiecutter templates that produce fully wired, policy-compliant service repos in under 10 minutes, eliminating copy-paste drift.
- TechDocs (docs-as-code) — Renders MkDocs Markdown stored in each service repo as a searchable documentation site inside the Backstage portal, keeping docs co-located with code.
- Argo CD (GitOps) — Watches the central GitOps repository and reconciles Kubernetes workloads on EKS, providing the deployment target for all scaffolded services.
- Vault (secrets management) — Provides per-service dynamic credentials fetched at pod startup via the Vault Agent Sidecar, integrated into every golden path template.
- GitHub Actions CI — Runs tests, Snyk security scans, and manifest rendering on every PR opened by the Scaffolder, enforcing quality gates before GitOps sync.
The Situation
This is a composite scenario assembled from patterns observed across multiple mid-size engineering organisations; all figures are illustrative and chosen to be internally consistent, not drawn from any single company's data.
The organisation is a 120-engineer B2B SaaS company selling a workflow-automation product to mid-market enterprises. Engineering is split into roughly 18 product squads, each owning two to five microservices, plus a six-person platform team responsible for the shared infrastructure. The production environment runs on AWS: EKS 1.29 for compute, RDS PostgreSQL for most persistence, and a mix of SQS and Kafka for async messaging. All services are deployed via Argo CD watching a central GitOps repository. The company had been on Kubernetes for three years, but the experience of *getting onto* Kubernetes had never been standardised.
The rest of this case study — the trade-offs, the scaling numbers and the build prompt — is available to LearnXops members. Create an account or see membership.
Topics: backstage, platform-engineering, developer-experience, golden-paths, kubernetes, internal-developer-platform