Running the meeting where it gets decided
Thirty minutes, six people, and a decision that has been drifting for two weeks. Two of them will not have read the document.
A meeting is a product, and most are badly designed
A meeting that ends in a decision
- Before: Name the decision
- First 10 min: Read the evidence
- Next 15 min: Resolve trade-offs
- Last 5 min: Owner and date
If people arrive unprepared, that is a flaw in your process rather than a character flaw in them. So fix it with structure.
Three rules carry almost all of it. State the decision in the invite as a sentence with a date in it. Give the first ten minutes to silent reading, which feels wasteful and saves the other twenty. End with a named owner and a date, or the meeting has not ended, it has been postponed while everyone pretends otherwise.
A meeting is useful when it changes what happens next. Preparation, a decision owner and a record matter more than a confident presentation.
Use your chosen course project throughout. The additional examples below are fictional practice cases; transfer the method to your own evidence.
Design the decision before the agenda
Write the choice as a sentence: approve a one-day investigation, choose one of two scope options, or stop a weak experiment. Identify who is authorised to make that choice and what information they need. If the agenda cannot name a choice, use a written update or a working session instead.
Give disagreement a place to land
Send a short brief with the options and uncertainties. During silent reading, let people mark assumptions rather than forcing them to interrupt a speaker. Group objections into evidence, constraints and preferences. Deal with the objection that changes the decision first, even when a less important point is easier to answer.
Close the loop while everyone is present
Read back the decision, owner, date and unresolved risk. Ask whether anyone heard something different. Record a revisit trigger for uncertainty that remains. A meeting with no decision can still be useful if it names the missing evidence and the person responsible for getting it by a specific time.
Make the decision observable
- Before: Brief and owner
- Read: Mark assumptions
- Discuss: Resolve the key risk
- Close: Decision and date
How Amazon handled it
Amazon
Amazon banned slide decks in decision meetings. The person proposing writes a six page narrative memo instead, and the first twenty minutes of the meeting are spent in silence, reading it.
Bezos's reasoning was that bullet points let you hide a weak argument behind a confident delivery, while prose forces you to connect your claims. The silent reading exists because “everyone read the doc” is never true, so the meeting stopped pretending.
Meeting design is product design. The format you choose determines the quality of the decision you get out of it.
Doing it with AI, and where it breaks
Draft the agenda and the decision statement, then ask the model for the three hardest questions someone could ask, and answer them in advance.
It produces generic objections about timeline and cost. Push it: “only ask questions that a person who has read the memo and disagrees with it would ask”. The generic ones cost you nothing, the specific ones are the meeting.
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.
Turn this decision brief into a 30-minute meeting plan. Name the decision owner, prerequisites, reading time, three objections likely to change the call, and the exact read-back at the end. Flag questions that cannot be answered from the brief. Do not schedule a presentation unless it serves the decision. Brief: [paste].
Before you use the output
- Make sure every agenda item serves the named choice
- check the owner is real
- preserve dissent in the final note.
Stuck? Try this next
Ask AI to play the busiest participant and identify what it could not decide after reading the brief. Fix the brief before rehearsing your speaking voice.
Keep a brief AI log: input used, useful output, what you checked, and what you rejected. The decision remains yours.
Build it
A run of show for a real thirty minute decision meeting, plus three hard questions and your answers.
The worked solution
Stuck, or want to compare?
There is a worked solution. Try the build first. Reading it before you have attempted anything is the fastest way to learn nothing.
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.