Teams are avoiding the internal platform and building their own tooling. What do you conclude?

What is being tested

Whether you read the behaviour as a product finding rather than as a discipline problem.

What to conclude

The platform is not good enough for their needs, and you now have the most valuable feedback available.

Teams under deadline pressure do not build their own tooling for entertainment. Doing so costs them time they do not have, which means the platform's cost to them exceeded that. Something specific is wrong.

The instinct to mandate adoption is exactly backwards: it removes the signal without addressing the cause, and it converts a product problem into a compliance problem that will be routed around less visibly.

What to investigate

1. What are they building, and why? Talk to them without judgement, or you will not get accurate information — and accurate information is the entire point.

2. Which of the usual causes applies?

• Missing capability. The platform does not do what they need. A roadmap item.
• Poor documentation. They could not work out how to use it. Documentation is a feature, and this is the cheapest possible fix.
• Too much friction. Onboarding takes a week; their own solution took two days.
• Too constraining. The platform enforces choices that do not fit their workload.
• Poor support. They tried, got stuck, could not get help, and gave up.
• Reliability. The platform broke and cost them an incident.

3. Is their need legitimate? Sometimes the workload genuinely differs and the right answer is an exception, recorded, with the team owning the consequences.

What to do

• Fix the cause, and make the fix visible to the teams who left.
• Reduce friction, particularly onboarding. Time to first successful use is the metric that predicts adoption.
• Provide an off-ramp with an honest cost: teams may leave the paved road and own their own pipeline, monitoring, on-call tooling and compliance evidence. That is not a punishment; it is the real cost, stated.
• Track exceptions as the roadmap. A pattern of teams leaving for the same reason identifies the next thing to build.

What not to do

Mandate it. That produces compliance without value, destroys the adoption signal, removes the incentive to improve, and generates resentment that outlasts the platform.

The framing

A platform that must be mandated is one that would not survive adoption on its merits. The paved road works because it is genuinely the easiest path, and compliance arrives as a side effect of convenience — the only form of governance that survives deadline pressure.

What a strong answer adds

That this is a self-service versus service desk question underneath. If teams are avoiding the platform because using it means filing a ticket and waiting, the problem is not the capability — it is the interaction model, and no amount of feature work fixes it.