I need a decision today and the team can't agree.

What to do when you need a decision today and the team can't agree

New Brunswick, New Jersey. Late September nineteen eighty-two. The Johnson and Johnson corporate offices.

Seven people in the Chicago metropolitan area have died after taking Tylenol Extra-Strength capsules laced with potassium cyanide. The first three deaths come within hours of each other on the morning of the twenty-ninth. The pattern is recognised before the day is out. The capsules were tampered with on the shelf, after they left the factory, somewhere between distribution and purchase.

The decision that arrives at the corporate offices in New Brunswick is not technically a decision the chief executive is required to make alone.

The board has not convened. The FBI investigation is still open — the contamination appears, on early evidence, to be local to the Chicago area, possibly the work of a single individual. Johnson and Johnson's own financial analysts are advising against a national recall. Pulling every bottle of Tylenol from every shelf in the country will cost roughly one hundred million dollars, in nineteen eighty-two money, and the company will not get that money back. The advice in the room is to wait. Localise. See what the FBI finds. Recall surgically when the data lands.

James Burke, the chief executive, recalls everything anyway. Within hours.

Before the data is complete. Before the advisers have finished advising. Before the board has had the meeting it would normally have on a decision of this size. He acts on the arithmetic he can already see — that the cost of being wrong on the side of caution is one hundred million dollars and the cost of being wrong on the side of waiting is whatever number of further deaths the company permits by hesitating — and he acts before the room has finished debating which side of the arithmetic to be on.

Johnson and Johnson's market share recovers within a year. The Tylenol recall is now the canonical case taught in business schools as the example of decisive crisis leadership.

What gets less attention is the structural point. The organisational machinery around Burke was designed, like every large corporate's machinery, to prevent exactly the kind of unilateral speed Burke uses. The committees, the legal review, the financial sign-off, the board governance — all of it is meant to slow the chief executive down, because most of the time slowing the chief executive down is the right thing to do. Burke ignored the machinery in the case where it would have produced the wrong answer.

Recognising which case you're in is the harder skill than the decision itself.

So let's go to the office and work through it.

"I need a decision today and the team can't agree."

The feeling is paralysed.

The clock is doing something. The team is doing something else. The thing the clock cares about and the thing the team cares about are not the same thing, and you're sitting between them.

First work out whether the team is stuck or deflecting

Two choices. Look the same on the surface. Take different moves.

When capable people have talked themselves into a knot

Choice one: the team is genuinely stuck. Capable people, working in good faith, who've talked the options into a knot. Each option has a defender. Each defender has a reasonable case. The conversation has been going long enough that everyone is now invested in their position rather than in the decision. Nobody is willing to be the one who concedes.

If that's the trap, change the cost of indecision. Email the team and say you'll execute Option A at five o'clock unless they come back before then with a unanimously agreed alternative. You haven't made the decision. You've turned the cost of further stalling from zero into a five-o'clock deadline, and you've forced the room to choose between the option you've named and a unified counter-proposal. Thaler's choice architecture, doing what it does. People who can't agree on the best option can almost always agree on the one they don't want you to pick by default.

When the indecision is quietly handing the call back to you

Choice two: the team is deflecting. They look stuck, but the indecision is functional. They have, somewhere along the way, recognised that as long as they don't agree, you'll be the one who has to call it. The disagreement is doing them a favour. It's letting them hand the decision back to you without admitting that's what they're doing.

If that's the trap, the choice-architecture move won't work — they'll let you go to five o'clock and execute Option A and treat that as a perfectly satisfactory outcome. Announce instead that you'll hire an external consultant tomorrow to make the decision if the team can't make it today. The team will find an answer within the hour. Not because consultants are bad. Because the prospect of having one brought in is an unambiguous signal that the team's contribution is no longer required, and very few people want that to be true about their own judgment. Schelling on credible commitment — the threat works because the threat is real, and the team knows you'll do it.

Diagnose before you intervene. The same-looking room takes opposite moves.

Fix who decides before you fix the decision

Three tools. The discipline is to fix the authority question before you fix the decision.

Name a single decider and end the authority fight

The first is

Rapid — the version of Decision Making Frameworks most managers know by name.

Bain and Company formalised rapid in the early two-thousands; the underlying discipline is older and broader, but rapid is the one most managers reach for.

The reason the tool exists is that the standard failure of a stalled meeting isn't that the decision is hard. It's that nobody has admitted, out loud, who gets to make it. The room debates the decision; what they're actually disagreeing about is the authority.

The unique insight is that single occupancy on Decide is the load-bearing constraint. Two-headed decisions are decisions that will collapse the moment the two heads disagree. Pick one Decider. Say the name.

What you get is a meeting where the conversation changes the moment the role assignment is explicit. The Recommender knows their job is to put the answer on the table. The Decider knows the call is theirs. The Input role knows their input is welcome but the decision is not theirs to block.

So. Five roles, assigned to people, not to the decision.

Recommend. The role that proposes the answer. Usually the person closest to the work.

Agree. The role that can veto, and only veto. Used sparingly — legal sign-off, regulatory approval, strategic alignment. Agree is not the same as approval; it's the small set of people whose disagreement is structurally fatal.

Perform. The role that will execute. Their input matters because they're the ones who'll discover whether the decision is operable.

Input. The roles whose advice is welcome but not binding. Advisers, stakeholders, the wider room.

Decide. The single role that calls it.

Settle for a decision the room can live with

The second is

Consensus Building.

The lineage runs through the Harvard Negotiation Project — Fisher and Ury, Getting to Yes, nineteen eighty-one — into the public-policy and group-process traditions of the nineteen-eighties and nineteen-nineties, where consensus had to be built across genuinely disagreeing parties without the option of a single decider.

The reason the tool exists is that most rooms confuse consensus with unanimity. They wait for everyone to want the same thing, most rooms never get there, and the decision gets made by the loudest voice or by attrition.

The unique insight is that consensus is not unanimity. It is the absence of a veto. The threshold is I can live with this — not this is the option I prefer.

What you get is a meeting that ends with a decision the room can carry, even when not everyone in the room would have picked it.

So. How to run it.

Frame. Open by naming what you're asking for. We're not looking for everyone to prefer this. We're looking for everyone to be able to live with it.

Hear. Let objections land before sorting them. Restate them back; check you've understood. The objector needs to feel heard before they can let go of the position.

Sort. A structural objection names a constraint the option fails on — illegal, unfunded, technically impossible, ethically off the table. A preferential objection names a different option the objector would have preferred. Structural objections are vetoes; the room reworks the option. Preferential objections are not vetoes; the room thanks the objector and proceeds.

The objectors who would prefer their preferences to count as vetoes will, predictably, not enjoy the conversation in which the distinction is named. That conversation is still the conversation that needs to happen.

Push the call down to the level it belongs

The third is

Delegation Frameworks.

Roy Williams developed the seven-level model in the nineteen-seventies, refined through subsequent leadership literature, including Stephen Covey's adaptations.

The reason the tool exists is that recurring meetings that won't decide are, surprisingly often, meetings at the wrong delegation level. Nobody is wrong about the merits. Everyone is wrong about whose call it is.

The unique insight is that when the level is wrong, the conversation is structurally undecidable. The team thinks they're at consult — the manager will decide and they're being asked for input. The manager thinks they're at agree — everyone has to concur. Or the team thinks they're at delegate and the manager thinks they're at tell and is waiting for someone to ask.

What you get is a way to surface the level-mismatch before it eats the meeting — and a way to push the decision down to where it can be made cleanly.

So. Seven levels, in increasing order of autonomy.

Tell. I make the decision and inform you.

Sell. I make the decision and explain why.

Consult. I ask for your input, then I decide.

Agree. We decide together; both of us must concur.

Advise. You decide; I'll offer advice if you ask.

Inquire. You decide; I'll ask afterward how it went.

Delegate. You decide; I trust your judgement and don't need to know.

Two questions to ask before the meeting. What level of delegation is this? And have I told the person? Most stalled meetings break the moment the level is named out loud and pushed to where it should sit. The team that couldn't decide could decide perfectly well at the right level. The meeting was the wrong forum.

That's the toolkit. One more story before we close.

The James Burke story we opened with is about acting on the arithmetic before the room has finished debating it. The story we close with is the same move at a different scale — someone who decided against the room, against the operationally-feasible response, because the operationally-feasible response would have taught the institution the wrong lesson.

A precedent: deciding against what the institution can absorb

Seattle. The twelfth of April, two thousand and eighteen. A Starbucks store in Philadelphia.

Two Black men — Rashon Nelson and Donte Robinson — are arrested for waiting to meet a business associate without having ordered. The video goes viral inside twenty-four hours. By the morning of the thirteenth, the brand is national news.

Inside Starbucks corporate, the leadership team is not in agreement about the response.

The standard playbook is on the table. Apology. Investigation. Revised staff guidance. Rolling racial-bias training delivered store-by-store over months, integrated into the existing operations cadence. It is defensible to the board, manageable for operations, and visible enough to the press to count as action. It is the response the institutional machinery is designed to produce.

The alternative, championed internally by Rosalind Brewer — Starbucks's chief operating officer — is structurally different.

Close every US company-operated store on a single day. Pull all one hundred and seventy-five thousand employees out of their shifts simultaneously. Run mandatory racial-bias education across the entire system in one afternoon. The lost sales are estimated at twelve million dollars. The operational disruption is severe.

Brewer's argument in the internal deliberations, reported across the Fortune and Marketing Interactive coverage of her tenure, was that an incremental training response would signal that the incident itself was incremental — absorbable by tuning the existing response machinery. The organisation needed the opposite signal. And signals cost money.

A spread-out training programme would have been cheaper, more compliant, and easier to sell to the board. It would also have taught the organisation that the incident was survivable without structural change. Brewer's position was that the response had to make it clear, internally and externally, that it was not.

The decision is announced on the seventeenth of April, executed on the twenty-ninth of May. Eight thousand stores closed. One hundred and seventy-five thousand employees in mandatory training on a single afternoon.

So. Burke acted before the room had finished debating, on the arithmetic he could already see. Brewer acted against the room's preferred response because the preferred response would have produced the wrong signal. Both decided against what the institution could comfortably absorb. Both did so deliberately, because the message in the absorption itself — this is the kind of incident we are tuned to handle — would have been the wrong message to send.

The deeper point: what your choice teaches the organisation

When you need a decision today and the team can't agree, the question is sometimes the one underneath the disagreement. What will the option you're about to pick teach the organisation about what kind of moment this is? A response calibrated to what the institution can comfortably absorb is a response that has already decided, in advance, that what happened was ordinary. Sometimes that's the right call. When it isn't, the cost of the disproportionate response is the message.

So. The question from this playbook.

What's the cost of the decision not being made by Friday — in pounds, in morale, in the opportunity you've stopped being able to take — and who in your next meeting is quietly comfortable paying it?