Building a Roadmap and Prioritisation in Practice

Premise — a roadmap is a table of "what we want to achieve," not "what to build and when"

A quarterly roadmap that lists only feature names is often a problem that predates prioritisation. A schedule of features makes a promise about what ships and when, but it leaves no record of the reasoning behind why those things were chosen. As a result, when resources run short partway through, no one can explain what should be cut.

This article assumes that the skeleton of your product strategy is already in place. If you introduce only a prioritisation method while the relationship between strategy and roadmap remains vague, you end up with initiatives that score highly but have nothing to do with the strategy sitting at the top of the list. The connection between strategy, vision, and roadmap is covered in The difference between product vision, strategy, and roadmap, and how they connect, so if the ordering is unclear, read that first. The other premise is that you have a list of candidate initiatives and a team able to produce at least rough estimates for each. Estimation accuracy varies greatly by business stage, and in the early phase of a new venture the Confidence value described next tends to be low across the board.

Procedure — five steps to prioritising with RICE scores

RICE is a method that Intercom's product team organised for internal use and then published. It computes a score per initiative from four factors: Reach, Impact, Confidence, and Effort.

1. Compile the candidate initiatives

List them regardless of origin — customer requests, support tickets, feedback from sales, internal ideas. At this stage, don't select or reject anything; focus only on aligning the level of granularity.

2. Estimate Reach

Estimate the number of users or uses that the initiative will affect within a set period (often a quarter). Whether you standardise the unit on "people" or on "sessions" needs to be decided by the team in advance.

3. Set Impact

Rate the strength of the effect on a single user on a graded scale. In the original scale Intercom published, the following guide values are used: 3 = massive, 2 = high, 1 = medium, 0.5 = low, 0.25 = minimal. You may adjust this scale to suit your organisation, but changing it midway destroys the comparability of scores.

4. Set Confidence

Express as a percentage how much basis there is for your Reach and Impact estimates. Set it high for estimates grounded in quantitative data, and low for estimates based on gut feel or a few opinions. The practice of estimating Confidence too generously is a prime example of the pitfalls discussed later.

5. Estimate Effort and calculate the score

Estimate the effort required for implementation, validation, and launch in a unit such as person-months, and calculate the score with the following formula.

RICE = (Reach × Impact × Confidence) ÷ Effort

Worked example — RICE scoring in numbers

The following is an illustrative example, not real figures. Suppose a SaaS product scored three candidate initiatives.

Initiative A  Simplify onboarding
  Reach 1,200 people/quarter · Impact 2 · Confidence 80% · Effort 3 person-months
  RICE = (1,200 × 2 × 0.8) ÷ 3 = 640

Initiative B  Add a recommendation feature
  Reach 400 people/quarter · Impact 3 · Confidence 50% · Effort 4 person-months
  RICE = (400 × 3 × 0.5) ÷ 4 = 150

Initiative C  Finer-grained notification settings
  Reach 2,000 people/quarter · Impact 0.5 · Confidence 90% · Effort 1 person-month
  RICE = (2,000 × 0.5 × 0.9) ÷ 1 = 900

In this example the scores rank in the order C, A, B. This clearly illustrates a characteristic of RICE: an initiative with large reach and small effort tends to score highly even when the impact per person is small.

Check — is it acceptable to decide on score alone?

It is safer to avoid making the ranking of scores directly into the execution order of your backlog. That said, this does not mean RICE fails to work. RICE is strong when comparing initiatives of a similar nature; if you compare an exploratory 0-to-1 bet and a fine-grained improvement in the refinement stage within the same table, the meaning of the score breaks down. The former has little basis for Confidence to begin with, while the latter has a stable denominator for Reach.

Another limitation is how to handle initiatives that are strategically important yet come out with a low score. Initiatives to respond to a competitor's moves, or to prevent future technical debt, tend to be rated low on short-term Reach and Impact. A realistic approach is to treat the score purely as a starting point for comparison, and to leave the final judgement to the product's owner in light of the strategy.

Where teams stumble — common pitfalls in running RICE

If the criteria for Impact or Confidence differ from one scorer to the next, the absolute value of the score becomes meaningless. You need to align on the criteria before scoring. You also see cases where the unit of Reach is mixed across initiatives — people, sessions, revenue — in which case comparing scores at all does not even hold up.

It is not unusual to see Confidence inflated for the reason that "if we lower it, the initiative won't get through." This is a state of having lost sight of the purpose of scoring. There are also cases where, when scores end up neck and neck, the final ranking is decided by the opinion of the loudest stakeholder — in which case the scoring itself has become a hollow formality.

The next step

Rather than turning the top-scoring initiatives directly into backlog tickets, organise them as roadmap items with a one-sentence statement of the outcome you want to achieve. Reviewing them each quarter and reflecting on whether the actual Reach figures and your Impact estimates matched will improve the accuracy of future scoring. The metric design that underpins prioritisation is covered in Designing and operating a North Star metric, so do take a look.