How do we prevent pilots from stealing production SRE cycles?

Gate pilots in a lab namespace with quotas and egress limits, require on-call shadowing only after a promotion decision, and use CODEOWNERS so platform/security review changes before they hit shared infra.

What if we miss roadmap commitments because of the 10% pilot allocation?

Then you overcommitted. Make pilot capacity explicit in sprint planning. If leadership wants to reclaim it, they should publicly agree to defer a scope slice or accept risk to the innovation pipeline.

How do we keep spikes from turning into stealth migrations?

Timebox to two weeks, require an ADR on day one, and schedule a kill-or-promote decision in the calendar. No decision, no continuation. Enforce via labels and the weekly triage.

Isn’t this just more process?

It’s the minimum viable process to protect exploration. The rituals are short, the gates are binary, and the guardrails are code. Compared to failed hack weeks and ad-hoc spikes, it’s less overhead and more outcomes.

Culture · Oct 31, 2025 · 10 minute read

Innovation Time Without the Theater: The 85/10/5 Model That Survives Q4

You don’t need hackathons or vague “20% time.” You need guardrails, rituals, and metrics that let exploration breathe without blowing your delivery SLOs.

Back to all posts

Innovation Time Without the Theater: The 85/10/5 Model That Survives Q4

Key takeaways

Implementation checklist