When you have more good ideas than you can build, the hard part is sequencing them, and a single “which feature do you want?” poll hides more than it reveals. This template runs every candidate through the four MoSCoW levels, so you see the intensity of demand, then adds one forced “ship this first” vote to surface the sharpest priority.
You leave with a demand matrix, not a single winner: what’s broadly loved, what’s narrowly adored, and what people want most when they can’t have everything. A what’s-missing box captures the ideas outside your list, so the roadmap keeps growing from real demand.
The four levels
The template uses MoSCoW as its authors defined it, and the intro block teaches the levels to respondents before they answer, which is what makes the answers comparable across people.
| Level | Weight | What it means |
|---|---|---|
| 🔴 Must have | 3 | The release fails for you without it. |
| 🟠 Should have | 2 | Painful to miss, but there’s a workaround. |
| 🟡 Could have | 1 | Nice, and the first thing you’d drop if time ran out. |
| ⚪ Won’t have this time | 0 | Fine to leave for a later release. |
Two things about that last level are easy to get wrong, and both are written into the copy on purpose. “Won’t have” is scoped to this release, not forever, so nobody reads it as a rejection and refuses to pick it. And the framework only works if Musts are scarce: a survey where every candidate comes back red has told you nothing.
Demand, quantified
Sorting each candidate turns vague enthusiasm into a number you can rank. Each respondent even sees their own “feature appetite” tallied live, a small payoff that puts Subo’s scoring on display, one of the features the data shows almost nobody discovers.
Variants
- More candidates or categories: add candidate questions and group them under section headers.
- Weighted by segment: ask role or plan first and weight power-user demand differently.
- Public request board: make write-in requests visible to build a lightweight Canny-style board.
- Value vs. effort: clone twice to build a value/frequency matrix.