RICE, ICE & WSJF Priority Calculator

What this tool does

CM8-340 scores a backlog with four prioritisation models at once — RICE, ICE, WSJF and plain value over effort — and shows where they rank the same work differently. It then draws a cut line where your capacity actually runs out, which is the part most prioritisation exercises leave out and the part that decides what really happens.

Everything runs inside this single file: no account, no upload, no network request, which matters when the contents are your roadmap.

RICE

Reach × Impact × Confidence ÷ Effort. Designed for product work where you can count who a change touches.

Reach is the number of people, orders, tickets or transactions affected in one period, and it must be something you could point at in a report. This is where RICE goes wrong most often: reach counts people, not events. A change used by forty warehouse staff forty times a day has a reach of forty, and the frequency has to be carried by the impact score instead — the sample data includes exactly that trap.

Impact uses the standard five-point scale from 0.25 to 3. Confidence is how much you trust your own reach and impact figures, and it is the honest brake on large speculative items: a big idea at 50% confidence correctly halves.

RICE's blind spot is time. It has no way to express "this must happen before March", which is why a regulatory deadline affecting twelve people ranks near the bottom.

ICE

Impact × Confidence × Ease. The quickest to fill in and the crudest — three numbers, no units, no reach.

In this tool, ease is derived from the effort figure rather than asked for separately: the least effortful item in the set scores 10 and everything else scales against it. That is a deliberate decision. Asking for ease as a separate opinion, when effort has already been estimated, produces two numbers that disagree with each other and lets people move an item up by re-scoring it as "easy" without changing the estimate.

ICE rewards small items heavily. Used alone it produces a backlog of quick wins and no structural work, which is why it is best used for a first pass rather than a decision.

WSJF

Cost of delay ÷ job size, where cost of delay is business value plus time criticality plus risk reduction and opportunity enablement, each scored 1 to 10.

WSJF is the only model here that can express urgency separately from value, and that single property makes it the right choice more often than its reputation suggests. Time criticality asks how fast the value decays if you wait — a deadline, a season, a contract date, a competitor's move. Risk reduction and opportunity enablement capture what an item unlocks beyond its own benefit, which is the only place in any of these models where platform and infrastructure work can score at all.

All three components are needed or none: the tool blocks a partial set, because two out of three produces a cost of delay that is quietly too low and an item that sinks for no reason anybody will notice.

Reading the disagreement

The agreement column is the most useful output in the tool, and it is why all four models are calculated rather than one.

When the models agree, the item is genuinely well or badly placed and the ranking can be trusted. When they disagree sharply, the item has a property one model can see and another cannot:

• High on WSJF, low on RICE — urgent but narrow. A deadline, a compliance change, a contract date. WSJF is usually right, because the consequence of being late is not proportional to reach.
• High on RICE, low on WSJF — broad but not urgent. Genuinely valuable and genuinely able to wait a quarter, which is worth knowing when capacity is tight.
• High on ICE, low on the others — small. Do it in the gaps; do not let it displace something structural.

Treat a disagreement as a prompt for a conversation, not as an error to be resolved by picking the model that gives the answer you wanted.

The cut line

Set your capacity for the period, in the same effort unit as the rows, and the tool accumulates effort down the ranked list and marks where it runs out.

This is the single most valuable thing a prioritised backlog can show, and a sorted list never does it. A ranking tells people their item is eighth; a cut line tells them their item is not happening this quarter. Those are very different conversations, and the second one is the honest one. It is also considerably easier to have three months early than three months late.

Be conservative with capacity. Use the effort your team actually delivers in a period, not its theoretical availability — support, meetings, holidays, interruptions and the work that arrives unplanned routinely take a third or more.

How these models get gamed

Any scoring system that decides resource allocation will be optimised against, usually without anybody intending to cheat.

• Reach inflation. "Everybody" becomes the whole customer base rather than the people who actually encounter the thing. Insist that reach can be sourced.
• Impact drift. After a few rounds, everything is a 2 or a 3. If a third of the backlog is scoring 3, the scale has collapsed — rescore against a fixed example everybody recognises.
• Effort optimism. The denominator is the easiest number to move, and halving it doubles the score. Estimates from the person who wants the work done are worth checking.
• Confidence at 100. Reserved for measured things. If everything is 100%, the column has stopped doing its job.

The defence is not a better formula. It is that the numbers are visible, consistent and challengeable — which is what this tool is actually for.

Choosing a model

Use RICE when you can count reach and nothing has a hard date. Use WSJF when timing matters, when deadlines are external, or when platform and risk work keeps losing. Use ICE for a fast first pass over a long list. Use value ÷ effort when confidence is uniformly high and you want the simplest defensible order.

Choose before you score, not after. Scoring first and then selecting the model that produces the preferred answer is a real risk with a tool that calculates all four, and it is worth naming.

The formulas

FAQ

What units should effort be in? Whatever you already use, as long as every row uses the same one. Mixing person-days and story points in one backlog produces a ranking that means nothing at all.

Why is my compliance item near the bottom on RICE? Because RICE cannot see deadlines. Switch to WSJF, or accept that some work sits outside prioritisation entirely — statutory obligations are not competing for a place in the list, they are constraints on what is left.

Two items have nearly the same score. Then the model has not separated them, and the tool marks them as tied. Break it on something real: which has an owner, which unblocks something else, which the team will actually finish.

Should done items stay in the list? Yes, marked done — they drop out of the ranking and the cut line but keep the history. Items in progress deliberately stay in, so the capacity they are consuming is visible.

Can I use this for a non-software backlog? Yes. Reach works for anything with users or transactions, and WSJF was designed for exactly the mixed backlog most operational teams have.

Saving your work

Initiatives, settings and the report header are written to this browser's local storage as you type. Treat Export .json as the real save, which Import .json restores anywhere. Export CSV gives you the backlog for spreadsheet work. Reset asks twice, then erases everything. There is no undo.

Accuracy & disclaimer

Every model here multiplies and divides estimates, and the arithmetic lends the result a confidence the inputs have not earned. Reach is a forecast, impact is an opinion on a five-point scale, and effort is the number teams get wrong most consistently and always in the same direction. A score of 42.7 is not more precise than one of 39.1.

What these models are good at is forcing the same questions to be asked of everything, exposing the item everybody assumed was obvious, and making the cut line explicit. Where two models disagree sharply, that disagreement is more informative than either score — and the decision belongs to a person.