Roadies latest work pulls the rug out from under a comfortable industry myth: DIY Backstage projects dont fail because Backstage is a bad idea they fail because teams treat an IDP like another infra repo instead of shipping and running a product.
The immediate, punchy data point is simple and ugly: organisations that spun up bespoke Backstage portals without a clear product org and operating budget are now paying for it in stalled upgrades, plugin fragmentation, and a worse developer experience than they had before. Upgrades get postponed, plugins fork into incompatible variants, and the storefront meant to reduce tickets becomes a growing source of them. Roadie calls this trend "DIY Backstage is dying," and it shows up in upgrade lag, plugin fragmentation, and rising support load.
If that sounds familiar, the Google Cloud platform engineering research report should land like a confirmation rather than a contradiction. Their core finding: teams that explicitly treat their internal developer platform as a product roadmap, SLAs, user research and instrument pipelines with Four Keys (which operationalizes DORA metrics) see measurable gains in lead time and deployment frequency. The operative change isnt fancy tech; its coupling platform outputs (templates, workflows, scaffolding) to developer-facing telemetry so the platform team can see whether the golden path is actually used and delivering value.
That connection is the missing link for many DIY efforts. It's not enough to give developers a catalog of plugins; you must know whether the catalog reduces queue time, increases deployment frequency, or lowers change failure rate. Roadie's data shows many DIY Backstage projects lack that feedback loop; Google Cloud shows teams that instrument it win.
Two behaviors keep showing up across the community: platforms that ship opinionated golden-path templates and self-service pipelines reduce ticket-driven workflows; and the platform team that survives behaves like a product org prioritising UX improvements, measuring queue time and lead time, and iterating on the platform experience rather than endlessly adding automation scripts. Publications and community forums have shifted from debating plugin lists to focusing on which developer journeys to optimise and how to measure them.
Practical, executable differences separate a doomed DIY from a resilient one. Resilient teams do three concrete things well: (1) commit to a release cadence for Backstage and its plugins, backed by CI that runs compatibility checks; (2) collect Four Keys telemetry and expose it in dashboards that developers actually use; (3) treat scaffolding and generators as product features with usage SLIs (queue time, template adoption, rollback frequency). Skip any of those and youre running a brittle fork of Backstage with rising operational costs.
Heres my blunt take: if your platform team is not staffed and funded to run Backstage like a product with engineering capacity, UX work, and telemetry ownership you should stop rolling your own portal and either adopt a productised offering or sharply narrow your scope to a set of opinionated templates you can fully operate. The industry is moving away from tool-centric signals and toward outcome-centric platforms; continuing to build unoperated DIY portals is just deferring painful technical debt.
Two small linkable notes for teams who want to act: Google Clouds report is a direct primer on measuring IDP impact with Four Keys and DORA-style metrics (Google Cloud research: Measuring IDP impact with Four Keys and DORA metrics). If youre trying to reduce developer wait time, see the practical guidance on measuring queue time and building self-service workflows (Measure Queue Time in Your Internal Developer Platform and Build SelfService Workflows).
Prediction: the market will bifurcate. Well-funded central platform teams will keep DIY Backstage alive and advance it as a product; everyone else will either buy an opinionated IDP or standardise on golden-path scaffolding that they can actually operate. If your org is currently treating Backstage like a weekend project, this is the week to stop kidding yourselves.