RICE vs impact/effort?

Impact/effort is simpler and less wrong. RICE creates false precision.

How often to reprioritize?

Quarterly commitment, monthly review, weekly adjustment.

Should sales input the roadmap?

Themes yes, specific features no. Sales optimizes for the next deal; the roadmap has to optimize for the next 100.

Roadmap Prioritization: Frameworks That Actually Work

RICE, MoSCoW, and value-vs-effort matrices are the popular frameworks. The one that actually works at startup scale is simpler.

Roadmap Prioritization for Startups

Prioritization frameworks are seductive because they replace judgment with math. At startup scale, they usually produce a false sense of rigor. The real question is which one bet moves the company forward most.

Why frameworks fail at startup scale

RICE assigns numbers you don't know (reach, impact). Effort estimates are wrong by 2-3x. Averaging fake precision gives you fake confidence. The scoring produces the answer the scorer wanted.

The single-question test

"What's the one thing that, if we shipped it this quarter, would most change the trajectory of the company?" That thing goes first. Everything else waits. If the team can't agree on the answer, the strategy is unclear — that's the real problem.

The three-lane structure

Big bet (the trajectory-changer). Compounding investment (paying down debt, retention improvements). Support work (customer commitments, bug fixes). Roughly 60/25/15. Never more than one big bet at a time.

What to cut

Features requested by fewer than 3 customers. Rewrites without a shipping deadline. "Foundation" work that has no downstream feature depending on it. Everything on the roadmap should have a customer or metric attached.

Frequently asked questions

RICE vs impact/effort?: Impact/effort is simpler and less wrong. RICE creates false precision.
How often to reprioritize?: Quarterly commitment, monthly review, weekly adjustment.
Should sales input the roadmap?: Themes yes, specific features no. Sales optimizes for the next deal; the roadmap has to optimize for the next 100.