Every product or data team has more candidate machine learning projects than it has
engineers to build them. The RICE framework (Reach, Impact, Confidence,
Effort) turns that shortlist into a single comparable number:
RICE = (Reach × Impact × Confidence) ÷ Effort. Each bar in the scene is one
candidate project, positioned on a value-effort matrix — effort along one
axis, reach × impact × confidence along the other — with bar height showing the resulting
RICE score.
RICE was popularised by Intercom's product team as a lightweight alternative to gut-feel roadmapping — it forces every pitch to state its assumptions about reach and confidence out loud, not just its expected impact.
A 3D value-effort matrix plots candidate machine learning projects by effort and expected value, with bar height showing each project's RICE score — and a live roadmap budget shows which ones a team can actually fund this year.
RICE = (Reach × Impact × Confidence) ÷ Effort turns fuzzy pitches into one comparable number. Bar position shows the value-effort trade-off; bar height and color show the score and whether the roadmap budget can afford it.
Shape "Your Project" with the Reach, Impact, Confidence and Effort sliders and watch its rank change against six real-world ML candidates. Drag the team capacity slider to see which projects survive the cut for a 12-month roadmap.
RICE deliberately keeps Effort in the denominator so that a merely "good" project with tiny effort can outrank a "great" project that would eat a whole team's year — cheap wins compound.