Roadmaps are good at compounding work and terrible at bleeding. When a real gap shows up — revenue leaking, merchants blocked — routing it through sprint planning means the fix ships weeks after everyone agreed it mattered. Our pattern for those cases: the top business gap gets a strike team. Product, engineering, QA — a handful of people pulled from their lanes, one goal, authority to ship.
A recent example. We found a segment of merchants hitting a dead end during install — no payment method configured on their store platform, no way forward, silent drop-off. A strike team formed the same day; an alternative path into the app was in production the next. Not heroics. The team was small enough to skip alignment meetings and senior enough not to need them. The same pattern is now carrying a much bigger swing for us — same mechanics: one goal, one room, daily cadence.
The discipline is in the constraints. One strike team at a time — if everything is urgent, nothing is. It disbands when the gap closes; a standing "special team" is just a second roadmap with better PR. And the gap has to be measured in business terms, not engineering ones, or you're relabeling refactoring. Roadmap for the compounding work, strike team for the bleeding.