ICE is a prioritisation score: Impact × Confidence × Ease. Rate each dimension (e.g. 1–10), multiply, and rank. High-ICE items get done first. The framework forces you to separate three questions that people often collapse: How much does this matter? How sure are we we can pull it off? How hard is it? Impact alone ignores feasibility — you might chase a moon shot that never lands. Ease alone ignores value — you might fill the calendar with quick wins that don't move the needle. Confidence bridges the two: a high-impact, high-ease idea that you're unsure about may be a gamble; the same idea with high confidence is a clear bet. Multiplying (rather than adding) means that a zero on any dimension kills the score. No impact, no point. No confidence, too risky. No ease, too expensive. The discipline is to score honestly and to re-score as you learn. ICE doesn't make the decision; it structures the comparison so you can decide with clarity.
The framework is often attributed to Sean Ellis and growth teams: score growth ideas by impact, confidence, and ease, then run the top of the list. It generalises. Use it for roadmap items, initiative prioritisation, pipeline opportunities, or backlog. The limit is that the numbers are subjective. Two people can score the same item differently. The value is less in the absolute score and more in the conversation: why did you give that a 7? What would have to be true for it to be a 9? The framework surfaces assumptions and aligns the team on what "impact" and "ease" mean in context.
Section 2
How to See It
ICE shows up wherever people are ranking options with a simple score. The diagnostic: are they explicitly rating impact, confidence (or likelihood), and ease (or effort)? Are they multiplying to get a single rank? Look for the 1–10 (or 1–5) scale and the triple dimension. Growth teams often have an "ICE score" column in their experiment tracker. Product teams use it for backlog. Sales use it for opportunity prioritisation.
Business
You're seeing ICE when a growth team scores 20 experiment ideas before the quarter. Each idea gets Impact (how much would this move the key metric?), Confidence (how sure are we it will work?), Ease (how fast can we run it?). They multiply, sort, and the top five get resourced. The rest go to the backlog. The framework prevents "we'll do everything" and "we'll do whatever's easiest."
Technology
You're seeing ICE when engineering ranks tech-debt paydown vs features. Impact: how much does this improve reliability or velocity? Confidence: do we know the fix works? Ease: how many sprints? A high-impact, high-confidence, high-ease fix goes first. A low-ease refactor with uncertain impact gets deprioritised unless impact or confidence is exceptional.
Investing
You're seeing ICE when an investor ranks pipeline companies. Impact: upside if this works. Confidence: conviction in the team and thesis. Ease: how much work to diligence, support, and get to a decision? The framework doesn't replace judgment, but it makes the trade-off explicit. High-impact, low-confidence names need more work or a pass; high-ease, low-impact names are quick no's.
Markets
You're seeing ICE when a trader ranks trade ideas. Impact: expected return. Confidence: edge and conviction. Ease: liquidity, execution cost, position size. Not every idea is tradeable at scale; not every tradeable idea has edge. ICE forces the three into one rank so the best use of risk budget rises to the top.
Section 3
How to Use It
Decision filter
"When you have a list of options and limited resource, score each on Impact, Confidence, and Ease. Multiply. Rank. Do the top of the list first. Re-score when you learn something that changes Impact or Confidence."
As a founder
Use ICE for roadmap, initiatives, and experiments. Score impact on the metric that matters most — revenue, retention, activation. Score confidence from your data and domain knowledge; if you're guessing, say so (low score). Score ease in time or cost. The mistake is stacking the deck: giving everything high scores. The move is to differentiate. If everything is 8-8-8, the framework isn't helping. Force rank. The second mistake is never re-scoring. When an experiment fails, confidence in similar ideas drops; when you learn something new, impact or ease might change. Update the scores and re-rank.
As an investor
Use ICE to rank pipeline and capital allocation. Impact is upside in the base case. Confidence is your conviction in team and thesis. Ease is deal dynamics, diligence load, and relationship. The framework doesn't replace judgment — you still have to score — but it makes the trade-off explicit. High-impact, low-confidence names need more work or a smaller check. Low-impact, high-ease names are quick no's. Re-score after meetings and as you learn.
As a decision-maker
Use ICE when the team is stuck choosing between options. Have each option scored on the three dimensions. Compare. The conversation will reveal disagreement on impact or confidence — resolve that, then the rank is clear. The discipline is to not add; multiply. A zero on any dimension should zero out the option. That forces honesty about impact, confidence, or ease.
Common misapplication: Inflating scores to justify what you already want. If every pet project gets 9-9-9, ICE is theatre. The value is in differentiation. Calibrate: what does a 5 look like? What does a 9 look like? Use the scale so that the ranking reflects real trade-offs.
Second misapplication: Ignoring ease and only chasing impact. Moonshots have high impact and often low confidence and low ease. ICE will rank them lower unless confidence or ease improves. That's the point: don't pour resource into long shots when high-ICE options exist. Balance the portfolio; let ICE surface the quick wins that compound.
Section 4
The Mechanism
Section 5
Founders & Leaders in Action
John DoerrChair, Kleiner Perkins; author, Measure What Matters (OKRs)
Doerr popularised OKRs — objectives and key results — which force prioritisation and measurable impact. ICE and OKRs align: both require explicit criteria (impact, key results) and ranking. Doerr's discipline of "measure what matters" is the same spirit as scoring impact in ICE. Founders who use OKRs to set priorities often use a framework like ICE to rank initiatives within an objective. Score, multiply, execute the top.
Graham's advice to founders — "do things that don't scale," "make something people want," "focus on growth" — implies a prioritisation: do the thing with the highest leverage. ICE makes that explicit. Impact is leverage; ease is "don't scale" (do the manual thing first); confidence comes from talking to users. Graham doesn't name ICE, but the logic is the same: rank by what moves the needle and what you can do now.
Section 6
Visual Explanation
ICE: Impact × Confidence × Ease. Multiply to rank. A zero on any dimension zeros the score.
Section 7
Connected Models
ICE sits among prioritisation and decision frameworks. The models below either overlap in purpose (Eisenhower, 80/20), explain a dimension (impact, opportunity cost), or warn against misuse (impact bias).
Reinforces
80/20 Rule (Pareto)
Pareto says a minority of causes drive most results. ICE makes that operational: impact is "how much result does this cause?" High-impact items in ICE are often the Pareto few. The reinforcement: 80/20 says focus on the vital few; ICE scores so you can find them. Use 80/20 to think about impact; use ICE to score and rank.
Reinforces
Eisenhower Decision Matrix
Eisenhower sorts by urgency and importance. ICE sorts by impact, confidence, and ease. The reinforcement: both force explicit criteria instead of "we'll do everything." Eisenhower is binary (urgent/not, important/not); ICE is continuous. For backlog and experiments, ICE gives finer granularity. For inbox and daily triage, Eisenhower is faster.
Tension
Impact Bias
Impact bias is the tendency to overestimate the emotional impact of future outcomes. In ICE, impact can be overstated if you're not careful. The tension: the framework asks for impact, but we're bad at predicting it. Mitigate by using data and base rates, and by re-scoring after you learn. Don't let impact be the dimension you inflate.
Tension
Opportunity Cost
Opportunity cost is what you give up by choosing one option. ICE ranks options but doesn't explicitly model the cost of the next-best alternative. The tension: doing the top ICE item has opportunity cost (the second item). The discipline is to remember that ranking is relative — the real question is "top vs next best," not "top vs perfect."
Section 8
One Key Quote
"Score your growth ideas on Impact, Confidence, and Ease. Do the highest scores first."
— Sean Ellis, GrowthHackers
The framework in one sentence. The power is in the multiplication: it forces you to balance the three. You can't win on impact alone if confidence or ease is zero. The practitioner's discipline is to score honestly, differentiate between options, and re-score when new information arrives. The quote doesn't mention that — but the framework only works if you do it.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
ICE is a conversation starter. The number is less important than the alignment. When you score impact, you're defining what "matters" means. When you score confidence, you're surfacing assumptions. When you score ease, you're being honest about capacity. Use the framework to get the team on the same page. Then rank.
Differentiate or don't bother. If every item is 7-7-7, ICE isn't helping. Calibrate the scale. What's a 3 impact? What's a 9 ease? Force spread. The value is in the ranking, and the ranking only works if the scores reflect real differences.
Re-score when you learn. An experiment fails → confidence in similar ideas drops. A constraint lifts → ease improves. New data → impact might change. Update the scores and re-rank. ICE is not a one-time exercise; it's a living prioritisation.
Ease is a force multiplier. High-impact, high-confidence, low-ease ideas are important but expensive. High-impact, high-confidence, high-ease ideas are the quick wins. Don't ignore the easy ones. They compound. ICE surfaces them.
Section 10
Summary
ICE ranks options by Impact × Confidence × Ease. Use it to prioritise experiments, roadmap, or pipeline. Score honestly; multiply; do the top first. Re-score as you learn. Avoid inflating scores or chasing only impact — balance matters. Connected ideas include 80/20, Eisenhower matrix, opportunity cost, and marginal cost/benefit.
Ellis and Brown on growth tactics and prioritisation. ICE (and variants like RICE) is used to rank experiments. The book gives context for when and how to score.
Ries on build-measure-learn and prioritising learning. ICE aligns with prioritising experiments that have high expected learning (impact × confidence × ease of running).
McKeown on doing fewer things better. ICE is a tool for the "fewer things" — score, rank, do the top. Essentialism is the philosophy; ICE is one implementation.
Leads-to
Marginal Cost/Benefit
Marginal cost/benefit asks: what's the next unit of effort worth? ICE is a way to approximate that for discrete initiatives: impact is benefit, ease is inverse cost. The lead-to: once you have ICE scores, you're implicitly comparing marginal benefit and cost across options. Do the high-benefit, low-cost items first.
Leads-to
Scheduling & Prioritization
Scheduling and prioritisation is the act of ordering work in time and importance. ICE is a method: score on three dimensions, multiply, rank. The lead-to: ICE is one way to prioritise. It's not the only way, but it's a simple, repeatable way that works for experiments, backlog, and pipeline. Use it when you have a list and need an order.
