Prioritisation
RICE scoring sheet
Reach, Impact, Confidence and Effort columns with the score formula built in, defined confidence bands, and a column recording the assumption behind each input.
What it is
RICE scores an idea on four things. Reach is how many people it affects in a period, Impact is how much it changes things for each of them, Confidence is how far you trust those two numbers, and Effort is what it costs. Score is Reach times Impact times Confidence, divided by Effort, so an idea that helps a few people enormously can outrank one that helps everybody slightly.
The framework came out of Intercom, which needed a way to compare ideas arriving from support, sales and the team without the loudest voice winning. That is still the job it does best. The arithmetic is ordinary, and the value lies in being made to write a number where you would otherwise have written an adjective.
Confidence is the column that keeps it honest. It is the one place where a team admits a reach figure came from a guess, and a low confidence score pulls a favourite idea down the list in a way an argument rarely can.
When to use it
- You have more requests than the team can take and no agreed way to compare them.
- Two stakeholders each believe their item matters most and neither has stated why in numbers.
- You want to show your reasoning rather than your conclusion, so the order can be challenged on its inputs.
- A quarter is starting and you are choosing between fifteen candidates rather than three.
The columns
- Item
- Reach
- Impact
- Confidence
- Effort
- Score
- Reach assumption
- Impact assumption
- Owner
The template
| Item | Reach | Impact | Confidence | Effort | Score | Reach assumption | Impact assumption | Owner |
|---|---|---|---|---|---|---|---|---|
| Bulk seat import from CSV | 400 | 2 | 0.8 | 2 | 320 | 400 admins on accounts above 50 seats, from the seats table on 4 Aug | Removes the whole manual setup step, seen in 6 of 8 sessions | R. Adeyemi |
| Download invoice as PDF | 1200 | 0.5 | 1 | 0.5 | 1200 | 1200 billing contacts opened the invoice page last quarter | Small but certain, it is the top request in billing tickets | J. Whitaker |
| Rebuild settings navigation | 3000 | 0.25 | 0.5 | 3 | 125 | Every active workspace admin | We think findability is the problem, nobody has tested it | Unassigned |
Score is Reach times Impact times Confidence, divided by Effort. In the sheet that is =(B2*C2*D2)/E2, filled down.
What goes in each column
- Reach. How many people this affects in one quarter, counted from a real query. Write the query behind the number in the reach assumption column, because that is the figure people will challenge.
- Impact. How much it moves things for each person reached, on the fixed scale below.
- Confidence. How much you trust the two numbers to the left of it, as a band rather than a feeling.
- Effort. Person months from everyone involved, design and engineering together, rounded to the nearest half.
Impact scale
- 3 for massive
- 2 for high
- 1 for medium
- 0.5 for low
- 0.25 for minimal
Confidence bands
- 100% when you have data on reach, data on impact, and an engineering estimate
- 80% when one of those three is an informed estimate
- 50% when you are working from a strong argument and no measurement
Anything you would score below 50% is not a candidate yet, it is a question, and the honest move is to run a small piece of discovery rather than to put a number on it.
Reading the results
The scores are for sorting, not for deciding. Sort the sheet, then read the top ten rows out loud with the team and ask whether the order is defensible. Where it is not, the fault is almost always in the reach assumption or the effort estimate, so fix the input rather than overriding the score, and leave a note in the cell saying what changed and when.
Questions people ask
- What if I cannot get a reach number?
- Write the number you would defend and put the query behind it in the reach assumption column. A stated assumption can be corrected. A blank cell gets filled in by whoever argues hardest.
- Is RICE better than weighted scoring or MoSCoW?
- It is more useful when the ideas are comparable and you have some data. MoSCoW is faster for agreeing the scope of a release. Weighted scoring fits when your criteria are specific to your business rather than generic.
- Should the highest score always win?
- No. Sort by score, then read the top rows aloud with the team and ask whether the order is defensible. Where it is not, the input is usually wrong rather than the method.
- How precise should Effort be?
- Person months to the nearest half, counting everyone involved rather than engineering alone. False precision does more damage here than a rough number, because Effort is the divisor.