What ICE measures
ICE trades precision for speed. Impact asks how much this moves your key metric. Confidence forces you to admit what you don't know. Ease is the inverse of effort — how quickly you can ship it. Multiply all three.
The Subjectivity Trap
Without a framework, prioritization becomes politics — HiPPO, loudest client, recency bias. Any framework is better than intuition alone. ICE doesn't eliminate subjectivity, but it makes it visible: when you score, you see where you're being subjective.
Single Feature Scorer
Feature Comparison
ICE's Blind Spot: The Instagram example from the slides is instructive — "Close Friends" scores ICE 320 vs Reels' 189. Yet Instagram built Reels because it was existential against TikTok. ICE cannot capture strategic importance or audience reach. Use it for small teams and quick decisions under 10 features; upgrade to RICE when reach and effort precision matter.
The Industry Standard
RICE is Intercom's framework and the de-facto default for product teams of 5–30 people. It adds Reach (how many users are actually impacted, not just exposed) and moves Effort to the denominator — penalising big bets more sharply than ICE's multiplicative Ease.
Confidence is the Key Dimension
The most common RICE mistake: defaulting Confidence to 100%. Per Kahneman's Noise (2021), two experienced PMs scoring the same backlog will produce wildly different priorities. If you haven't validated with clients, your Confidence is 50%. Honest confidence scoring halves inflated scores and forces real conversations.
Single Feature Scorer
Feature Comparison
Spotify example from the slides: The "boring" Sleep Timer (R=10M, I=0.5, C=100%, E=2pw) scores RICE 2.5 versus the "exciting" AI-Powered DJ (R=8M, I=2, C=50%, E=12pw) at RICE 0.67 — 4× lower. RICE penalises ambitious bets with low confidence and high effort, which is exactly what you want in a resource-constrained roadmap.
Weighted Shortest Job First
WSJF comes from SAFe / Lean Agile. Its key insight: the cost of NOT doing something is often more important than its benefit. It captures urgency and time-criticality that RICE misses entirely — e.g. a shelf-life feature that must ship before summer.
When to use WSJF
Use WSJF when there's a strong time dimension: regulatory deadlines, seasonal windows, competitive threats with a clock on them. It shines in SAFe environments with 30+ people. For smaller teams, RICE is simpler and usually sufficient.
Single Feature Scorer
Cost of Delay Components (1–10 each)
Feature Comparison
Not a Scoring Model — a Classification
MoSCoW doesn't rank features numerically. It forces a binary conversation: without this, is the release broken? Its power lies in the Won't Have category — explicitly committing to what you're NOT building this cycle, which removes the ambiguity that kills scopes.
Best Use: Scope Negotiation
Use MoSCoW when negotiating scope with a client or stakeholder. After RICE-scoring your backlog, translate the top items to Must/Should for the presentation. The "Won't Have (this time)" framing is softer than "rejected" and preserves the relationship while holding the line.
Add Features to Classify
Classification Board
Must Have
Should Have
Could Have
Won't Have
Combine frameworks: Use RICE to rank your backlog numerically, then translate the ranked list into MoSCoW categories to present the roadmap to your client. The frameworks solve different problems — RICE is for internal prioritization, MoSCoW is for stakeholder communication.
The Core Insight
No single framework is universally best. The right choice depends on your team size, the number of features, and whether there's a time dimension. Misapplying WSJF in a 3-person startup is as harmful as using ICE when strategic reach matters.
Start with RICE
RICE is your default. Master it first. The Confidence dimension alone — if used honestly — will transform your prioritization by separating assumptions from validated data. ICE is a training wheel; WSJF is a specialist tool; MoSCoW is a communication layer on top of the others.
Framework Selection Cheat Sheet
Common Traps — RICE Edition
Gaming the Reach
"All 1000 users will see the button" ≠ "all 1000 users will be meaningfully affected." Reach = users whose outcomes actually change, not eyeballs.
Impact Inflation
Everything is "massive" when the PM is excited. Use the scale honestly: 3 = game-changer, 0.5 = nice to have. If everything is 3, the scale is broken.
Ignoring Confidence
The most common mistake. People skip it or default to 100%. If you haven't validated with clients, your confidence is 50% by default.
Effort Sandbagging
Engineers pad estimates; PMs underestimate them. Use historical data: how long did similar features actually take? Always anchor to past velocity.
The Strategy Hierarchy — Context for Prioritization
Every backlog decision should be traceable up the hierarchy. If it isn't, you're not prioritizing — you're just filling time.
→ This is where your framework scores live. The backlog. But every score should reflect the strategy above it.