Economy research · Standard sources
Build vs buy an internal developer platform for a 200-engineer company
Completed August 1, 2026
Build vs Buy an Internal Developer Platform at 200 Engineers
Bottom line
For a 200-engineer company, the best-supported default is to buy or hybridize: start with a commercial or managed IDP (Port, Cortex, OpsLevel, or managed Backstage) and extend it with custom golden paths and integrations. A pure build on Backstage OSS only makes sense if you already have a funded platform team of 3+ senior engineers, can wait 12-18 months, and need deep customization or air-gapped operation. Hybrid is often the right answer: a managed core plus custom edges.
Why the choice is three-way, not two-way
- Build on an open-source framework, usually Backstage. Backstage is a framework, not a finished product; you own hosting, PostgreSQL, TechDocs storage, auth, monitoring, and integrations.
- Buy a commercial IDP (Port, Cortex, OpsLevel, Harness IDP, Atlassian Compass) or orchestrator (Humanitec). Commercial products ship a functional portal in days rather than months.
- Hybrid: buy a managed base (Port, Roadie-managed Backstage, Spotify Portal) and extend with custom code. Practitioner sources often call this the correct answer.
Decision table
| Path | Best when | Avoid when | Typical cost and time |
|---|---|---|---|
| Build on Backstage OSS | Deep customization; air-gapped or compliance control; existing platform team able to retain 3+ senior engineers | Speed matters; small or no platform team; no appetite for operating 5-8 infrastructure components | About $150K-$500K Year 1; 6-18 months to production; 2-3+ engineers ongoing |
| Buy commercial IDP (Port, Cortex, OpsLevel, Harness) | Fast time-to-value; standard toolchain; SaaS acceptable; small platform team | Air-gapped environments; data residency constraints; deep custom workflows outside vendor APIs | About $12K-$300K per year; 2-6 weeks to a basic portal; 1-3+ months integration |
| Hybrid (managed base + custom extensions) | Most 200-engineer companies; custom golden paths needed but managed core desired | Vendor integration limits block critical workflows | About $24-$65 per developer per month plus integration effort; 1-3 months to value |
Cost and timing ranges
Directional figures from 2025-2026 sources:
- Backstage OSS: $0 license, but 2-3 senior engineers full-time equals roughly $300K-$450K per year at $150K loaded salary, before custom plugins.
- Custom build benchmarks: Year 1 $90K-$1M+; 3-year TCO $200K-$2M+; 6-18 months to production. One worked example was $395K growing to $602K over 3 years, with a 13-month timeline.
- Self-hosted Backstage: $150K-$500K Year 1; vendor-adjacent sources say 3-6 months, but realistic production timelines are longer once CI/CD, cloud, secrets, and compliance integrations are included.
- Commercial SaaS: $12K-$300K per year; 2-6 weeks to a functional portal, but full integration and adoption take longer.
- Broader platform team at scale: about 10 platform engineers at 100 product engineers costs roughly $2.77M per year all-in; platform spend runs $11K-$18K per developer, with about 69% spent on salaries.
DORA 2024 caveat: IDPs correlate with +6% productivity and a dedicated platform team helps, but IDPs were also associated with an 8% decrease in throughput and reduced change stability when mandated across the full lifecycle. Heavy governance gates can slow delivery.
Reconciling the staffing contradiction
The research contains an apparent contradiction: a Backstage build typically needs 2-3 senior engineers, while platform-team ratios suggest 17-25 engineers at 200 headcount. The likely resolution is that the 17-25 figure describes the entire platform engineering organization (cloud, CI/CD, observability, security, internal tools), not the IDP core. Build-vs-buy models should separate:
- Core IDP build and maintenance: 2-3+ engineers.
- Integrations and golden-path content: often 1-2 additional engineers.
- Entire platform function: 10-25 engineers depending on scope and company needs.
Treat these as separate budget lines; do not use the 17-25 figure as a pure IDP cost.
Key risks and hidden costs
- Integration glue is where rollouts quietly fall behind schedule; most sources plan for a demo portal in 2-6 weeks but production adoption takes longer.
- Practitioner sources say custom-portal maintenance is underestimated by about 3x; one source claims roughly $180K per year of hidden platform toil, but such figures are single-source and not audited.
- SSO, RBAC, audit logs, data residency, and multi-tenancy are often unaddressed. Backstage has no native multi-tenancy; Port has workspace-based isolation; SaaS-only options may fail air-gapped requirements.
- The 249 active Backstage plugins are an ecosystem strength, but plugin quality and maintenance vary; every customized integration becomes an internal product you own.
- An IDP that is not adopted is wasted regardless of build or buy. Budget for champions, training, and documentation.
Recommendations
- Start with a 2-4 week vendor pilot for one or two commercial or managed candidates.
- Define done as at least 3 golden paths adopted by real teams, not an empty portal.
- List hard constraints first: air-gapped, data residency, SSO, RBAC, and audit requirements.
- Build a TCO model with explicit assumptions for integration, maintenance (about 3x license), adoption, and opportunity cost of senior engineers.
- Prefer hybrid unless a pure build is already justified by existing team and requirements.
- Only choose pure Backstage if you can fund and retain 3+ senior platform engineers for at least 18 months.
Evidence quality and gaps
The strongest evidence is DORA 2024 and Backstage ecosystem data. Most cost figures come from vendor-adjacent or practitioner blogs and should be treated as directional, not audited. There is no public benchmark isolating exactly 200-engineer companies, and commercial pricing is largely sales-led and unpublished. No source reliably separates the productivity opportunity cost of a build at this scale from the subscription cost of a buy.
Key sources
- DORA 2024 State of DevOps report
- Backstage plugin registry (backstage.io/plugins)
- CIOPages IDP Buyer's Guide (March 2026)
- Humanitec, Roadie, Port, Cortex, and Spotify Portal vendor documentation
- Practitioner analyses: Clouditive, LocalOps, ZopDev, developerportalcost.com, platformengineeringcost.com, and Eltherion
Assumptions and caveats
- Costs assume loaded engineer cost around $150K-$250K per year; adjust to your geography.
- The report assumes 200 engineers are product or application engineers served by the IDP.
- Production means adopted by real teams with CI/CD, cloud, secrets, and compliance integrated, not a demo portal.