Your internal platform has been live for a year. Six of twenty teams use it; the rest built their own pipelines. The CTO proposes mandating it. What is your advice?

Advise against the mandate, and say why in terms of evidence

A mandate converts a usability problem into a compliance problem. It will produce adoption numbers and not the benefit — teams will meet the letter of the mandate while routing around the parts that do not work, and the platform team will lose the signal that tells them what to build. Fourteen teams choosing to build their own pipelines is data, and it is being discarded.

Find out what the fourteen actually did

Interview them, and look at what they built rather than what they say. The findings usually fall into four groups:

• They did not know it existed, or believed it was not ready. A communication and onboarding problem, and the cheapest to fix.
• They tried it and bounced at a specific step. This is the most valuable group — ask exactly where. It is usually one missing capability, not a general dissatisfaction.
• They have a genuine requirement the platform does not meet — a language, a compliance boundary, a deployment target. Legitimate, and the question is whether it is worth supporting.
• They enjoy building pipelines. Real, and the only group where a nudge is the right instrument.

The diagnostic question

Is following the paved road easier than not following it? If onboarding requires a request form and a three-day wait, while writing your own pipeline takes an afternoon, the platform is losing on merit and every team is behaving rationally.

Adoption is a usability outcome. A platform that wins on merit does not need a mandate; a platform that needs one will not deliver the benefit.

What to do instead

1. Fix the top bounce reason. One capability usually unblocks several teams.
2. Reduce the cost of moving. Do the migration for the two largest holdouts yourself. Reference implementations from respected teams move others.
3. Make the value legible — deployment frequency, mean time to restore, security findings, cost per service, compared between paved-road and self-built teams. If the platform is genuinely better, this is your argument.
4. Instrument the platform. Track where onboarding is abandoned and which defaults are overridden.

Where a mandate is legitimate

Not for the platform as a whole, but for specific outcomes: signed artifacts, no long-lived cloud credentials, evidence of change control, vulnerability SLAs. Mandate the outcome; let the paved road be the easiest way to achieve it. Teams that meet it another way are complying, and their approach is worth studying.

If you are overruled

Ask for the mandate to come with a deadline that is a year out, a migration budget, and a commitment to fix the top three bounce reasons first. And keep measuring departure — teams that comply and then quietly build a shadow pipeline are telling you the mandate did not work.