A platform engineering playbook for VPs of Platform delivering an IDP that engineers actually use.
Engineers route around it.
Most IDPs fail because they are built like infrastructure projects, not like products.
The product discipline that wins is to define the golden path first — the most common workflow, end to end, that the IDP will optimize.
Working primitives over half-built ones.
Pick the single most common workflow. Probably 'deploy a new microservice' or 'add a new database.' Map every step engineers do today.
Roll out to additional teams, one team at a time, with paired support from the platform team for the first week per team. The pairing surfaces friction and feeds the next iteration.
Add the second golden path (provisioning, secret management, or runbook automation are common). Harden the first golden path based on adoption feedback.
Pick the single most common workflow.
Roll out to additional teams, one team at a time, with paired support from the platform team for the first week per team.
Add the second golden path (provisioning, secret management, or runbook automation are common).
If your IDP has been 'almost ready' for nine months, the answer is not more features.
First golden path can be delivered by 4 to 6 engineers. Scaling adoption needs a PM and 2 more engineers. We have shipped MVPs with smaller teams when paired with our embedded engagement.
Mix. Buy for capabilities with mature commercial offerings (CI/CD platforms, secret management). Build for opinionated workflow that is unique to your engineering culture.
We re-scope to a single golden path, sunset the unused capabilities, and rebuild adoption from a narrow base. We have done this rescue four times in 2025-26.
