Useful Consulting Frameworks
All frameworks

Innovation Decide

MoSCoW and RICE Prioritization

MoSCoW and RICE are two ways to choose what to build first: must, should, could, or won't, or a score from reach, impact, confidence, and effort.

The questionOf the things we could build, which must ship in this timebox, and which return the most for the effort?

Must, Should, Could, Won't this time. RICE is Reach, Impact, Confidence, and Effort.

Formula · Dai Clegg (MoSCoW); Intercom, Sean McBride (RICE), 1994 and 2018

Must

  • Generate the invoice
  • Apply tax
  • Post to the ledger

Should

  • Credit notes
  • Email the PDF

Could

  • Branded templates
  • Payment links

Won't

  • Multi-entity consolidation
MoSCoW buckets for one timeboxed release. RICE scores the same list when you need a rank, not only a cut.

When to use it

  • A release date is fixed and the list of requests will not fit, so some items have to be dropped in public.
  • Ranking is done by whoever argues longest, and you need a shared cut (MoSCoW) or a shared score (RICE).
  • Several ideas look important, and you need to separate a legal or viability must-have from a high-scoring nice-to-have.
  • You want a number that records confidence, so a vivid story with thin evidence does not beat a boring item with data.

When to avoid it

  • Dependencies dominate. A Must that rests on a Won't is a broken board. Sort dependencies first, or the labels will lie.
  • Effort cannot be estimated even roughly, and Reach is a guess. RICE then ranks confidence in storytelling.
  • The list is three items and the owner already knows the order. Do not add a framework to a conversation that is finished.
  • MoSCoW is being used as a wish list with no timebox. Without a date, everything becomes Must.

Overview

MoSCoW cuts a timeboxed list into must, should, could, and won’t. RICE ranks items by (Reach × Impact × Confidence) / Effort. Together they support the choice of what ships in this window, and in what order, when requests exceed capacity.

MoSCoW fails when Must is a synonym for requested. RICE fails when Effort is gamed, Reach is a market-size slide, or Confidence is always 100 percent. Use MoSCoW to protect a viable release, use RICE to order what is left, and do not let a score override a legal Must or a dependency.

MoSCoW categories and RICE inputs

Must

The minimum usable set. If this item is missing, there is no point shipping on the date. Legal, safety, and "the solution is not viable without it" belong here. A painful workaround means it is not Must. DSDM's rule of thumb is that Must effort should stay around 60 percent of the timebox, so there is room to absorb misses.

Should

Important, and painful to leave out, but the release is still viable without it. A workaround exists. Should is the first place to cut when the timebox slips.

Could

Wanted if time remains. DSDM treats Could as the main contingency pool, often around 20 percent of effort. If Could is empty, there is no buffer, and any overrun hits Must.

Won't (this time)

Agreed out of this timebox, in the open, so it stops returning as a whispered Must. It can come back next time. Won't is a decision, not a parking lot you refuse to look at.

Reach

How many people this will affect in a defined period, such as customers per quarter. Use the same period for every item. A percentage of users is fine if the base is stated.

Impact

How much it moves the chosen goal for each person reached. Intercom's scale is 3 massive, 2 high, 1 medium, 0.5 low, 0.25 minimal. Impact is the most subjective input, so write the goal it is scored against.

Confidence

How sure you are of the Reach, Impact, and Effort figures. Intercom uses 100 percent high, 80 percent medium, 50 percent low. Low confidence should hurt the score. That is the point.

Effort

Person-months, or another consistent unit of work, using whole numbers and a floor of half a month in Intercom's version. Effort is the only input in the denominator. Understating it is the usual way to game the rank.

What it is not

  • A single method. MoSCoW is a cut against a date. RICE is a score for ranking. Use MoSCoW to protect a viable release, RICE to order the Shoulds and Coulds, or to rank when there is no hard date.
  • A substitute for a strategy. If the goal that Impact is scored against is unspecified, RICE ranks charisma.
  • High-medium-low with extra letters. MoSCoW only works if Must is scarce and Won't is real.

The steps

  1. Fix the timebox and the goal

    Name the date, the capacity, and the outcome you are scoring Impact against. MoSCoW without a date is a wish list. RICE without a goal is a debate about adjectives.

  2. Cut with MoSCoW, and keep Must scarce

    Ask of each item whether the release is pointless without it. Move anything with a workaround to Should or Could. Put explicit Won'ts on the board. Check that Must effort is well under capacity, around 60 percent if you follow DSDM.

  3. Score the contested items with RICE

    Estimate Reach, Impact, Confidence, and Effort on the same scales for every remaining Should and Could, and for Musts only if you need a build order. Write the figures down before you multiply.

  4. Rank, then sanity-check dependencies and politics

    Sort by RICE score. Pull items that a Must depends on. Do not let a high score smuggle in a Won't. Record the exceptions and why.

  5. Re-cut when the timebox moves

    If work overruns, cut Could first, then Should. Recalculate RICE only when Reach or Effort has actually changed, not to justify a pet item.

The formula

RICE = (Reach × Impact × Confidence) / Effort

Estimated impact per unit of effort, scaled by how sure you are of the inputs.

  • Reach = number of people affected in a stated period, such as customers per quarter.
  • Impact = 3 (massive), 2 (high), 1 (medium), 0.5 (low), or 0.25 (minimal), scored against one stated goal.
  • Confidence = 100% (high), 80% (medium), or 50% (low), applied to the other estimates.
  • Effort = person-months, or another consistent work unit, typically in whole numbers with a 0.5 floor.

In practice

A billing-product team has a twelve-week release and nineteen items. Fourteen are labeled Must in the first workshop. Must effort adds up to 94 percent of the team's capacity. DSDM's 60 percent test fails immediately. The consultant asks, for each Must, whether the release would be cancelled without it. Generate invoice, apply tax, and post to the ledger stay Must. Credit notes move to Should because finance can issue them by hand for a quarter. Templates, payment links, and a dashboard move to Could. Multi-entity consolidation is Won't this time, written on the board so it stops arriving in Slack.

The remaining Shoulds and Coulds are scored with RICE against "paying customers who send at least one invoice per week," over one quarter. Email-the-PDF: Reach 1,800, Impact 2, Confidence 80 percent, Effort 1 person-month, score 2,880. Branded templates: Reach 400, Impact 0.5, Confidence 50 percent, Effort 1, score 100. Payment links: Reach 900, Impact 2, Confidence 50 percent, Effort 2, score 450. Email-the-PDF is pulled into the release as Should. Templates stay Could and then drop when a tax-rule change consumes the buffer. The 94 percent Must list would have slipped the date and still missed the PDF.

What you leave with

  • A timeboxed list with every item in Must, Should, Could, or Won't, and Must effort under capacity.
  • RICE scores for the contested items, with the four inputs visible, not the headline number alone.
  • A stated goal and period that every Impact and Reach figure uses.
  • Explicit Won'ts, so they stop returning as unofficial Musts.
  • A rule for what gets cut first if the timebox slips.

Failure points

Everything is Must

Must becomes a synonym for "someone asked." The timebox is then a fiction, and MoSCoW adds no information. Apply the cancel-the-release test and the 60 percent effort check.

Effort sandbagged or vanished

Teams lower Effort to raise the RICE score, or leave it out. Because Effort is the denominator, it is the easiest input to game. Estimate it in the same unit, with the same person, across the list.

Reach as total addressable market

A huge Reach figure is applied to a feature that only current users will see, or that only a sliver will use. Reach has to be the people this item will actually affect in the period.

Confidence set to 100 percent by default

Every item is scored as certain. The Confidence term then does no work, and thin stories rank with measured ones. Force 50 percent unless there is evidence.

Score used as the decision

The highest RICE number ships even when it depends on a Won't, breaks a Must, or fails a legal test. The formula is an input to the cut, not a substitute for it.

The failures above are usually Framing: the wrong problem was chosen before this framework was applied.

Origin

Dai Clegg developed MoSCoW in 1994 while at Oracle, for timeboxed RAD work, and published it with Richard Barker in Case Method Fast-Track: A RAD Approach. The DSDM Consortium, now the Agile Business Consortium, took it as the standard way to cut scope when time and cost are fixed. The o's in the name are there to make the word pronounceable. Won't means will not have this time, not never.

RICE is a later, separate scoring method from Intercom. Sean McBride, then a product manager there, published it in January 2018 as "RICE: Simple prioritization for product managers." Reach, impact, and confidence sit in the numerator; effort sits in the denominator. Nobody owns MoSCoW. RICE is not trademarked. Intercom's post is the source text for the factor definitions and the impact scale.

Further reading

Last updated