Your team decides before the ending.

One real company’s case, argued in writing before anyone knows how it ended. In an hour you see how your team reasons; a second session a few weeks later shows whether it moved.

Run the first sessionSee what the comparison shows
Watch a case room, start to finish · 3 minWatch the course together, start to finish · 3 minThe presentation for whoever approves it
decidenda.com · Multiplayer

What do you want to run?

One case together

One link · one format · you open the results

The course together

3–8 weeks · a first and a final decision

One session today. The comparison in a few weeks.

  1. About an hour

    Your team answers one case

    Each person argues the call in writing before the discussion and before knowing the ending. One link, nothing to install.

  2. Right after

    You see how they reason

    How the room split, the reasons people chose to show, and the step the group handled thinnest, when one stands out. Never anyone’s score.

  3. 2–4 weeks later

    The next session shows whether it moved

    The same group on a new case. Its results compare the people who answered both, step by step, in points out of 20 with the range the change likely falls in.

Compared with the previous session

Example

Answered in both: 9

  • Focusno clear change11 → 12 of 20 · +1, likely −1 to +2
  • Relevant evidenceno clear change10 → 10 of 20 · 0, likely −1 to +2
  • Alternativesstronger8 → 11 of 20 · +3, likely +1 to +4
  • Manage uncertaintystronger8 → 10 of 20 · +2, likely +1 to +4
  • Execute and learnno clear change11 → 11 of 20 · 0, likely −2 to +1

Where the same people stood on each step the first time and the second, and each person’s own change, averaged, in whole points out of 20. “Likely” is where the group’s average change most probably lies (a 90% range): it is wider when fewer people answered or when they moved in different directions. A step is called stronger or thinner only when it moved by 1.5 points or more and the whole range is above or below zero. Points are not percentages: +2 of 20 does not mean “10% better”, and has nothing to do with the 1% in a payback model. The points come from the AI reviews of each answer, not from a teacher’s or a host’s marking. The same people on two decisions, step by step. The cases differ and there is no control group: an observation, not proof that the practice caused it.

Choose by the time you have.

30–90 minutes

Solve one case together.

Use a live room for an offsite, class, leadership meeting or one-hour workshop.

Create a case room

Host sign-in required

  1. 01

    Choose a case and a format

    The whole case, or one FRAME step: the questions where it decides the most.

  2. 02

    Share one public link

    People can open the invitation before they sign in.

  3. 03

    Everyone answers first

    Each answer goes into the room when it is sent. Nobody sees anyone else’s until the results open.

  4. 04

    Open the results when ready

    How the room split, and the answers people chose to show, without names. On screen or printed.

3–8 weeks

Take the course together.

The course chapter by chapter, between a first and a final decision on real cases.

See the course plan

Public preview · no sign-in

  1. 01

    Choose 3, 6 or 8 weeks

  2. 02

    Add your people

    Paste emails, choose a start date and send the public invitation.

  3. 03

    The chapters, at each person’s pace

    Between the two decisions each person takes the chapters: a fictional company’s story, its methods, a test and a memo with feedback.

  4. 04

    Close with the report

    The group’s first and final decisions compared, never anyone’s work.

First and final decisions compared

Example

First decision: Rolls-Royce · Final decision: Klarna · Made both: 12

  • Focusno clear change11 → 12 of 20 · +1, likely 0 to +2
  • Relevant evidenceno clear change10 → 10 of 20 · 0, likely −1 to +2
  • Alternativesstronger8 → 11 of 20 · +3, likely +1 to +4
  • Manage uncertaintystronger8 → 11 of 20 · +3, likely +2 to +4
  • Execute and learnno clear change11 → 12 of 20 · +1, likely 0 to +2

Where the same people stood on each step the first time and the second, and each person’s own change, averaged, in whole points out of 20. “Likely” is where the group’s average change most probably lies (a 90% range): it is wider when fewer people answered or when they moved in different directions. A step is called stronger or thinner only when it moved by 1.5 points or more and the whole range is above or below zero. Points are not percentages: +2 of 20 does not mean “10% better”, and has nothing to do with the 1% in a payback model. The points come from the AI reviews of each answer, not from a teacher’s or a host’s marking. The same people on two decisions, step by step. The cases differ and there is no control group: an observation, not proof that the practice caused it.

One FRAME language inside both formats.

FRAME gives every case, review and debrief the same five questions.

F

Focus

Name the decision and its limits.

R

Relevant evidence

Use facts that could change the call.

A

Alternatives

Compare a serious second path.

M

Manage uncertainty

Name the risk, trigger and fallback.

E

Execute and learn

Commit and say when to revisit.

The person leading reads the work. The group never sees a name.

Each participant keeps

  • Their course work: nobody reads a course’s decisions or chapters
  • An anonymous answer, when the room allows one
  • An inbox we write to only for sign-in and for the group they joined

The organiser sees

  • Who joined, and how many questions each sent
  • In a room, every answer and its review, with the name, unless the room allows anonymous answers
  • The split, and the comparison of two sessions, from five people

No individual ranking. No use for hiring, promotion, dismissal or compensation.

Feedle Inc runs decidenda · answers are reviewed by an OpenAI model in the United States, which never receives a name or an address · data stored in the United States, or in the UK, the EU or Kazakhstan on request, when your rules require it · Data processing agreement

Fit the format to the team you already serve.

01

Corporate L&D

One decision process for manager development and leadership offsites.

See the team formats
02

Consulting firms

Build a repeatable path from case evidence to a committed recommendation.

See the team formats
03

Corporate universities

Add a complete decision-practice track to the academy you already run.

See the team formats
04

Executive education

Prepare executives individually, then use the room for the hard trade-offs.

See the team formats
05

Business schools

Written feedback on every student’s call, while grading stays with faculty.

See the team formats

Improving the judgement of 10 middle managers by just 1% is worth $100,000 a year.

The model

middle managers

decided by each one a year

1%

more return on that money

$100,000

a year

Put in your own figure: the team, budget and contracts one manager decides on.

The 1% is a model, not something decidenda measures. The points out of 20 a series or a course shows are not percentages, and do not go into this sum.

Why 1% is cautious: across 1,048 major company decisions, a decision process in the top quarter went with 6.9 percentage points more return on investment than one in the bottom quarter. It is a correlation from managers’ own accounts, not an experiment. That process is what you practise here. Lovallo and Sibony, McKinsey Quarterly, 2010

Personal practice is free. A team pays once or by the month.

No minimum team size · 150 players per room

Personal

$0

  • Complete FRAME course
  • Full case library
  • Written feedback
  • Personal decision workbook
Start practising

Pay once

$149a case room

$599a course

Your first case room and the course in 3 weeks are free.

  • One case for the whole group, up to 150 players
  • The course in 6 or 8 weeks for one group of up to 30
  • The same results and report as the subscription
Run your first room free

Team

$49per participant a month

The first month is free, no card needed.

  • Every case room and course, without limit
  • Series and programmes
  • Results without anyone’s score
Start with your team

Questions about multiplayer.

Can the host decide when a room starts and ends?
Yes. Run it in the session, with players in a lobby until you start, or before it, with answers open at once. Players can join until you open the results. A room takes 150 players besides the people leading it.
Can I run only part of a case?
Yes. A FRAME step format asks only the questions where that step decides the most, and keeps the rest of the case closed until the results.
Is this an employee assessment?
No. decidenda is for development and deliberate practice. Learning signals must not be used for hiring, promotion, dismissal or compensation decisions.

Choose the right multiplayer format

We will recommend a case room or a prepared cohort programme.