Skip to content
Drill75 minutesLesson 18 of 24

Which one is worth a quarter?

Five real problems, one team, twelve weeks. The founder wants all five and has said so in front of the team.

Confidence is the honest column

RICE is Reach times Impact times Confidence, divided by Effort. Reach and impact are estimates dressed as numbers. Confidence is where you record how much you actually know.

Which is why teams that adopt RICE and get nothing from it are the ones who set every confidence to 100 percent. The framework does not make the decision. It makes your assumptions visible so the argument is about the right thing.

FEATUREREACHIMPACTCONFIDENCEEFFORTSCOREFix bank sync8,200390%211,070Dark mode4,0001100%31,333AI assistant9,000340%81,350Export to CSV900290%11,620Confidence is the honest column. The AI assistant scores badly because nobody knows if it works.
The AI assistant has the biggest reach and loses, because nobody knows whether it works.
Work through it

A prioritisation score helps a team discuss assumptions consistently. It does not remove judgment or make a weak estimate objective by giving it decimal places.

Use your chosen course project throughout. The additional examples below are fictional practice cases; transfer the method to your own evidence.

Choose comparable inputs

For a RICE exercise, use reach over the same period, a shared impact scale, confidence as a fraction, and effort in the same unit. Define each scale before scoring. If one proposal uses annual reach and another monthly reach, the ranking is an artifact of the spreadsheet rather than a meaningful comparison.

Attach evidence to confidence

Confidence should reflect the support behind reach and impact, not how much someone likes the idea. Record the source and date next to each estimate. Keep dependencies, mandatory work, and strategic constraints visible outside the score; not every decision belongs in the same ranking exercise.

Test ranking stability

Vary uncertain inputs and see whether the order changes. If two options swap after a small adjustment, treat them as close and discuss evidence or a smaller test. The next best action may be research that improves confidence rather than immediate delivery of the current highest-scoring feature.

The visual field guide

Make the score auditable

  1. Reach: Same period
  2. Impact: Shared scale
  3. Confidence: Evidence quality
  4. Effort: Same work unit
Make the score auditable. Apply this sequence to your own project; it is a conceptual guide, not measured data.
Try the concept

How much confidence changes the choice?

Interactive teaching example · All numbers and tickets are fictional.

AB0100A: 50 · B: 64 · Option B ranks higher

A = 100 reach × 2 impact × 0.5 confidence ÷ 2 person-months. B = 80 × 1 × 0.8 ÷ 1 = 64. A needs confidence above 64% to lead. Evidence must justify that change.

Doing it with AI, and where it breaks

The move

Build the scoring sheet, then ask which of these scores is not supported by the evidence I have described.

The trap

It will accept your numbers as given and do arithmetic on them, which flatters you. Make it attack the inputs rather than compute the output.

Your AI workbench

Start with your own notes or clearly labelled practice data. Remove private details before sharing. Replace the placeholders, run the prompt in your chosen AI tool, and keep the output beside its source.

Prompt worth stealing

Audit this prioritisation table [paste]. Check periods, scales, confidence fractions, effort units, dependencies, and source quality. Recalculate each score transparently and show which plausible input changes reverse the ranking. Do not choose confidence values just to support my preferred idea.

Before you use the output

  • All proposals use consistent units
  • Confidence has an evidence rationale
  • Dependencies and close rankings remain visible
Stuck? Try this next

Ask AI to argue for the lower-ranked option using only supplied facts. This can expose a missing constraint, but it cannot manufacture evidence for that option.

Keep a brief AI log: input used, useful output, what you checked, and what you rejected. The decision remains yours.

Build it

The artefact

A scored shortlist where the confidence column is the honest one.

0/5

Checkpoint

If you did the build, these take two minutes. If you cannot answer one of them, that is the part to go back to.

Moving on marks this lesson complete. Finish the build first, it is the part that counts.

Previous lessonNumbers without a data team