The Plan Everyone Agreed to And Nobody Believed

The billing migration was due 30 April. It went live on 6 June.

What made the meeting afterwards strange was the absence of shock. Maya went round the table asking when each person first thought the April date was unrealistic. The earliest answer was the day the date was set. The latest was mid-March. Six people, six private doubts, one unanimous show of hands in a fifty-minute meeting on 12 February.

Nobody lied. Everybody agreed. The plan was dead on arrival, and it took fifteen weeks for that to become visible.

Back to the room on 12 February

Maya presented the timeline and asked whether it was doable. Rehan said the engineering side was tight but workable. Sofia said the vendor piece depended on their sandbox being ready, and if it slipped, everything slipped. Dan said nothing. Two others nodded.

Then Maya asked the question that ended the meeting: does everyone agree?

Six hands. Meeting over in forty-one minutes, ten under the booking.

Jerry Harvey named this failure in 1974, and the story he used to explain it has nothing to do with work. On a July afternoon in Coleman, Texas, with the temperature at 104 degrees, his father-in-law said, "Let's get in the car and go to Abilene." It was 53 miles in an unair-conditioned 1958 Buick. Harvey didn't want to go and said nothing because his wife sounded keen. His wife didn't want to go either. Neither did the man who suggested it. Four people drove 106 miles through dust to eat a bad meal, and every one of them privately opposed the trip.

Harvey's point in the Organizational Dynamics article was that organizations are far worse at managing agreement than at managing conflict. The 12 February meeting was a trip to Abilene with a Gantt chart.

The signal sitting in plain view

Sofia's sentence was the signal. The vendor piece depends on their sandbox, and if it slips, everything slips.

It was recorded in the notes. Maya's reply was, "Noted, let's flag it if it becomes an issue." Everyone moved on, and the flag was never raised, because raising it later would have meant contradicting a plan that had already been agreed upon by six people, including herself.

That's the trap. Once the show of hands happens, the cost of dissent goes up rather than down. Solomon Asch measured how much. In his line-judging experiments, participants gave an answer they could see was wrong on about 37% of the critical trials, and roughly three in four went along with the group at least once. These were people denying the evidence of their own eyes about the length of a line, with no career consequences at stake.

Asch also found the variable that breaks it. Add one other person who disagrees with the majority, and conformity collapses. Unanimity is what does the work. The moment it cracks, people say what they think.

The Tuesday it was still fixable

Fourteen March. The vendor confirmed their sandbox would be ready "late April, possibly early May."

Rehan read the email, worked out that this made 30 April arithmetically impossible, and forwarded it to Maya with the line "FYI, might be tight." Maya was in back-to-back interviews and replied with a thumbs up.

That was the last off-ramp. Six weeks of runway still existed. A different plan, a phased cutover, or a public reset of the date were all available on 14 March at a cost of one hour and some mild embarrassment.

The word "might" did the damage. Rehan already knew it was impossible. He wrote "might be tight" because the alternative was announcing that the plan everyone had agreed to was broken, and he had no evidence that anyone else thought so. He was Asch's lone participant, staring at a line, hoping someone else would speak first.

What you'd do differently

Let’s talk about how the whole thing can be done differently and more efficiently

• Get the numbers before you get the room. Send the plan out 24 hours ahead with one question: on a scale of one to five, how confident are you in this date? Collect the answers privately. Confidence scores make dissent arithmetic and take the courage out of it. A room averaging 2.4 tells you everything the show of hands hides.
• Run the pre-mortem before you ask for commitment. The framing that works: it's now 30 April, we missed it, write down why. Give people four minutes and silence. You will get the vendor sandbox in the first thirty seconds.
• Assign the disagreement rather than hoping for it. Name one person before the meeting whose job is to argue against the plan. Rotate it. The Asch finding says one visible dissenter is enough to unlock the rest of the room, and appointing the role means nobody has to volunteer for the social cost.
• Stop asking whether everyone agrees. Ask what would have to be true for this date to hold. The first question asks for permission to end the meeting. The second asks for the assumptions, which is where the risk lives.

And when someone writes "might be tight," treat it as the loudest thing anyone said that week. Reply with a question that costs them nothing: what would you put the odds at?

Your version of this

There's a date on your plan right now that at least one person on your team privately believes will not hold.

You could find out today. Send them the number question, promise you won't quote anyone, and read the scores before you defend the date.

Maya sends it now before every plan lands. The billing migration's replacement project came in on the second date she proposed, which was six weeks later than the first one she almost announced.