Your Developer Platform Failed Before Anyone Logged In

Most Internal Developer Platforms fail before a single engineer logs in.

The architecture was not wrong, and the portal framework was not the problem. The team building it never treated it as a product. They treated it as infrastructure with a UI slapped on top.

I have seen this pattern repeatedly. A platform team spends 18 months building something. They have Backstage running, a service catalog with 200 entries, golden paths documented in Confluence. Then adoption sits at 11%.

So they add more features.

The problem was never features. It was that nobody interviewed the developers who would use it. Nobody defined what friction they were actually solving. Nobody had a product manager. The "golden path" was golden for the architects who designed it, not the engineers who have to ship at 11pm on a Friday.

A platform is a product. It has customers. Those customers can choose to ignore it, and they will, the moment your internal tooling creates more cognitive load than just writing a bash script and moving on.

Adoption is the metric, not components deployed, not catalog entries, not uptime.

If your developers are building workarounds around your platform, the platform is the problem.

The hardest conversation a platform team can have is admitting they built something nobody asked for. Most never have it.