> "Two teams shipped the same internal dashboard, six months apart, in different stacks — nobody knew the other was building it."
^pain
A collective synthesis pain. Different teams unknowingly build the same model, library, or analysis. Discovery fails; people recreate from scratch. The cost is the duplicate engineering effort plus the divergence — now there are two versions to maintain, neither canonical.
## Discovery questions
- "Pick a function on your team - reporting, segmentation, analytics. How many parallel versions exist across the org?"
- "When was the last time you discovered another team had already built what yours was building?"
^discovery-questions
## Examples
- Spotify created Backstage and "Golden Paths" because teams were independently building their own deployment scripts, dashboards, and service boilerplates.[^1]
- Netflix's "Paved Road" platform work explicitly addresses teams building parallel tooling and infrastructure stacks by providing a supported standard path.[^2]
- Google's internal developer-productivity initiatives (Blaze/Bazel, Borg, Spanner as shared platforms) arose from fragmented build systems and duplicate infrastructure.[^3]
- Meta centralised React, GraphQL, and internal deployment tools after many product teams were independently solving similar UI and data-fetching problems.[^4]
- Uber's move toward a centralised microservice platform and service catalog was driven by proliferation of independently-created services with overlapping functionality.[^5]
[^1]: https://backstage.io
[^2]: https://netflixtechblog.com/full-cycle-developers-at-netflix-a08c31f83249
[^3]: https://research.google
[^4]: https://engineering.fb.com
[^5]: https://www.uber.com/us/en/blog/up-portable-microservices-ready-for-the-cloud/