Before we start, can we write down the decision we need to leave with? I want to make sure we are not just sharing updates.
The launch approach. We need the group comfortable with the plan before we commit to September.
Just checking: is the group making the decision, or are we giving Victor enough input to make it? Those are different meetings.
Sales can work with September. The risk is that the date moves after we have already set expectations with the accounts.
Customer Success has three risks from the first cohort. I can bring them into the decision, but I do not think the whole team needs to approve the rollout.
I have put the options, risks and proposed owner in a one-page brief. Can everyone add corrections there? Victor, could you make the call by 15:30?
Yes, please. If we still need another meeting after this, let us be clear about the decision it will make and who needs to attend.
The alignment meeting began seven minutes late because the previous alignment meeting had ended six minutes late and two participants needed a minute to relocate their sense of purpose.
There were eleven people in the room, twelve if you counted the conference display, which repeatedly announced that it could not detect a signal while showing everyone the signal.
Maya had organised the meeting because Enterprise Launch had developed three dates, four definitions of βreadyβ and a commercial commitment whose exact wording lived in Julesβs memory. Victor wanted alignment. Tom wanted a feasibility decision. Elena wanted the migration risks acknowledged. Leon wanted to know whether the launch experience was staged or merely inconsistent.
Nora attended because the invite said Final alignment in bold.
Maya shared the agenda:
- Context.
- Perspectives.
- Alignment.
- Next steps.
There was no item called Decision. This omission had survived three agenda reviews.
For forty minutes, everyone contributed useful information. This is what made the meeting dangerous. Bad meetings are easy to stop. A meeting full of intelligent context can continue until the heat death of the company.
Agreement is not the only form of alignment
Teams often use alignment to mean agreement, shared understanding, permission, commitment, awareness or the temporary absence of visible conflict.
Those states are different.
People can understand a decision and disagree with it. They can agree with a direction but misunderstand their responsibility. They can commit to execution while preferring another option. They can attend a meeting, say nothing and later discover that their silence was recorded as enthusiastic consent.
Our meeting wanted all six states simultaneously.
Jules explained why September mattered commercially. Tom explained why full enterprise migration by September created reliability risk. Elena described which customers could tolerate a staged rollout. Leon showed two experience options. Bea added activation data from a smaller segment. Every contribution improved the picture.
Nobody knew who would choose.
At minute fifty-three, Maya asked whether the group preferred staged launch A or broad launch B.
βI can support either,β someone said, which is the business equivalent of placing your steering wheel gently on the table.
βI would like more confidence,β said someone else, without specifying what evidence would create it.
Victor asked whether there was a recommendation. Maya said yes. Tom said technically yes, but with a condition. Jules said commercially yes, if the date held. Elena said operationally yes, if the first cohort changed.
The recommendation had become a family of recommendations sharing a surname.
Collaboration without decision design
Inviting more people can improve a decision when they contribute distinct evidence, expertise or consequences. It can also distribute ownership until nobody can move without recreating the room.
Atlassianβs DACI framework separates Driver, Approver, Contributors and Informed participants. The useful mechanism is not the acronym. It is the explicit distinction between voice and final authority. Contributors shape a decision; one approver makes it; informed people do not need to attend the making of it.
The Enterprise Launch meeting had nine contributors, no declared approver and two informed participants who had been invited βfor visibilityβ but were now understandably offering opinions because visibility had come with a microphone.
Maya paused the discussion and opened a one-page brief. Victor was the approver. Maya was the driver. Tom, Elena, Jules and Leon were contributors. Nora and several functional leads were informed. The decision was whether to stage the enterprise launch by account risk. The deadline was today. The revisit trigger was a failed reliability threshold in the first cohort.
Nothing about the underlying problem changed.
The social geometry changed immediately.
People stopped editing every sentence toward universal comfort and began testing the recommendation against their expertise.
Meeting β pre-alignment β final alignment β launch decision sync
- Driver: Maya β prepares options and closes open evidence
- Approver: Victor β chooses by 15:30
- Contributors: Engineering, Sales, Success and Design
- Informed: everyone whose calendar can now be returned
Disagreement is not a delivery defect
Many alignment rituals are attempts to remove disagreement before action. This is expensive and sometimes impossible.
The goal should be high-quality input, an understood decision and genuine commitmentβnot identical private beliefs.
In Amazonβs 2016 shareholder letter, Jeff Bezos describes βdisagree and commitβ as a way to preserve decision velocity when sincere disagreement remains, including leaders committing to a teamβs direction. The important half is not βcommitβ used as a command. It is that disagreement is expressed candidly, weighed, and followed by real support rather than passive sabotage.
That requires trust and a visible decider. Without them, βdisagree and commitβ becomes βyour concerns have been noted attractively.β
Tom disagreed with the proposed cohort size. He believed it exposed too many accounts before the migration path was proven. Jules believed a smaller cohort weakened the September commercial story. Both positions were reasonable because they optimised different risks.
Victor chose the smaller cohort with a pre-agreed expansion threshold. Jules disagreed and committed, after clarifying exactly what Sales could say. Tom committed and took responsibility for the threshold instrumentation. Elena documented which customers entered first.
The room did not achieve consensus.
It achieved a decision people understood well enough to execute.
Write the decision, not the meeting
Meetings disappear quickly. Their interpretations do not.
Martin Fowlerβs description of Architecture Decision Records focuses on short documents capturing a decision, its context and consequences. Although ADRs come from software architecture, the wider habit is valuable for product work: writing forces differences into view and leaves a durable explanation of why the organisation chose one path.
Mayaβs launch record contained:
- the decision in one sentence;
- the context and objective;
- serious options considered;
- the people who contributed and the approver;
- consequences and owner actions;
- confidence and the signal that would trigger review.
It did not contain a transcript. A decision record is not meeting minutes with better posture. It preserves the choice and reasoning, not every conversational branch that failed to become one.
This reduced a familiar form of alignment debt. Two weeks later, a regional lead asked why the rollout was staged. Instead of convening the original cast, Maya shared the record. The lead still disagreed, but could see the commercial, technical and customer constraints. They suggested a better expansion signal, which the team adopted without reopening the whole decision.
The meeting budget
Noraβs joke about charging the second meeting to the first was not entirely a joke.
Coordination has a cost. Eleven people for ninety minutes is more than a calendar rectangle. It is concentrated organisational attention, preparation and context switching. The cost may be worthwhile for consequential choices. Treating every stakeholder as a required attendee is not respect; it is an inability to design participation.
Use meetings for interaction that benefits from simultaneity: resolving a live disagreement, interpreting ambiguous evidence, exploring trade-offs or making a decision with the required people present.
Use documents for context people can absorb asynchronously. Use one-to-one conversations when psychological safety or expertise makes the group format worse. Inform people after a decision when their role is awareness, not contribution.
And if a meeting is for decision-making, place the decision in the invitation:
We will decide X. Maya drives. Victor approves. These four people contribute. Read the brief. If evidence Y is missing, we postpone; otherwise we choose today.
This wording feels severe only because most meeting descriptions ask for an hour without promising an outcome.
The Enterprise Launch meeting ended at 15:19, eleven minutes early. Nobody knew what to do with the time. Tom remained seated in case another meeting tried to claim it.
Maya published the decision record. Jules updated the commercial narrative. Elena prepared the cohort. Victor messaged Nora: βAligned.β
Nora replied: βDo you mean agreed, understood, committed, informed or temporarily quiet?β
Victor added a checkmark reaction and returned to work.
My take
Alignment is not the absence of disagreement. It is enough shared context, clear authority and explicit commitment for coordinated action.
Design the decision before designing the meeting. Name who contributes, who decides, what evidence matters and what will reopen the choice. Otherwise alignment expands to fill every calendar available.
Questions to take back to your team
- What decision will your next alignment meeting actually make?
- Is each attendee contributing, approving or simply being informed?
- Which disagreement are you trying to understand, and which are you trying to erase?
- Can somebody explain the final decision without replaying the meeting?
- What evidence or event would legitimately cause the decision to be revisited?
Open the next alignment invite and add one sentence naming the decision and approver. If you cannot, the meeting may not need more participants. It may need a purpose.