RICE, ICE, MoSCoW and Kano. Not a dump of frameworks — a guide to which one to reach for and when none of them help.
The common frameworks:
RICE — Reach × Impact × Confidence ÷ Effort. Best with data. ICE — Impact × Confidence × Ease. RICE's faster cousin, for early stage. MoSCoW — Must / Should / Could / Won't. Best for a fixed deadline. Kano — sorts features into basics, performance and delighters. Best for understanding satisfaction.
Each answers a different question.
Because frameworks are decision aids, not decision makers, and using the wrong one produces confident nonsense.
RICE with invented numbers is just your intuition with arithmetic on top — and it is more persuasive than intuition, which makes it worse. Knowing when a framework does not apply is the actual skill.
Use ICE early. Pre-product-market fit you have no reach data. ICE takes two minutes per item and is honest about being rough.
Use RICE once you have usage data. 'This affects 2,400 monthly active users' is a real input. Reach is the term that separates RICE from ICE and it needs real numbers to mean anything.
Use MoSCoW for a fixed date. Shipping for a conference on the 14th: what must exist, what should, what could, what explicitly will not. It handles the deadline constraint the others ignore.
Use Kano to understand complaints. Basics are invisible when present and infuriating when absent (login working). Performance features scale with quality (speed). Delighters are unexpected. If your product is missing a basic, no amount of delighters compensates — and RICE will not tell you that.
Use none of them when: you have fewer than about twenty customers. Then the answer is 'do what the customers you have are asking for', and a scoring exercise across eight items is theatre.
First-time founders often adopt RICE, invent the numbers, and treat the output as objective. Confidence is the term that should catch this, and it is the one people always set to 80% regardless.
The second mistake: using a framework to avoid a judgement call. Sometimes the right answer is 'the CEO thinks this matters and is willing to be accountable for it'. That is legitimate, and dressing it up in a score is worse than saying it.
Fixed date → MoSCoW. Have usage data → RICE. Early and rough → ICE. Understanding satisfaction → Kano.
If you are guessing, set confidence low. A low-confidence item scoring high is a signal to go and find out, not to build.
Reach must come from analytics. Otherwise use ICE and stop pretending.
If the ranking feels obviously wrong, it probably is. Find the input that is lying rather than overriding silently.
Talk to them and do what they ask. A scoring matrix at that scale is procrastination with a spreadsheet.
Once you have more good ideas than capacity — usually after product-market fit, not before.
Pre-PMF with a handful of customers. Also avoid when the decision is genuinely strategic — a framework cannot tell you whether to enter a new market.
Apply this to your own startup in My Full Journey (free account).