When the work gets complicated enough that nobody can hold it in their head, five lines on a page will do more than another meeting.
Bottom line. When a problem sprawls and nothing is moving, write down the five things that have to be true and start. Keep it short enough to run in a minute. Include only the steps that matter when they are missed. Have someone other than the doer read it out loud. Give it an owner who is allowed to delete things. Most of what stalls a team is not a knowledge problem, and a checklist is the cheapest tool for turning knowing into doing.
The plane that was too complicated to fly
In 1935 Boeing brought a new bomber to Wright Field for an Army competition. It was faster than anything in the fleet, carried more, and flew farther. On its demonstration flight it stalled on takeoff and crashed. The aircraft turned out to be sound. The pilot had missed a step, and the pilot was the Army’s chief of flight testing.
Newspapers described the airplane as too much airplane for one man to fly.
The machine had outgrown the person operating it, and more training was not going to fix that. The man at the controls already had as much as anyone alive. So a group of pilots wrote out the checks for takeoff, flight, landing, and taxi, short enough to fit on an index card. With that card, the aircraft went on to fly more than a million miles without an accident, and the Army bought thousands of them.
What a checklist does
Checklists have nothing to do with intelligence. Say that first when you bring one to a skeptical team.
A checklist does not tell you something you did not know. Every engineer on your team knows to check whether the migration is reversible. The list exists because knowing and doing come apart under load, and load is the normal condition of any job worth having.
Three good things happen when you write one.
The complicated becomes finite. A launch has a hundred moving parts and maybe six that can hurt you. Working out which six takes real thought, and you get most of the benefit before anyone runs the list.
The work becomes shared. A checklist is an object other people can hold. It survives the person who wrote it, the reorg, the vacation, and the engineer who joined last week. Most institutional knowledge lives in somebody’s head until they leave. This is the cheapest form of it that persists.
Something starts. Complicated problems produce meetings, and meetings produce more meetings. A list of five concrete steps produces a first step with a name next to it. It turns a discussion into an action.
Keeping it simple, which is the hard part
Every checklist wants to grow. Each surprise adds an item, nobody removes one, and a year later you have a forty-line document that gets signed rather than read.
A few rules keep it short.
Fit it on one screen and run it in about a minute. Longer than that and people start skipping. That is worse than having no list, because now you believe the work was verified.
Include only the steps that matter when they are missed. Everything else is documentation, and documentation belongs somewhere else.
Do not write instructions. A checklist confirms that something happened. The doing lives in people’s hands and in the runbook. Pilots configure the aircraft from memory and then use the checklist to verify a handful of critical items, which is why the list stays short.
Have someone other than the doer read it. Pilots use challenge and response: one reads the item, the other verifies and answers. The person who did the work is the worst person to confirm it, which is the same reason we ask someone else to review the code.
Give it an owner who is allowed to delete things. Without that, the list dies of its own weight.
Why now
The work keeps getting more complicated. More systems, more integrations, more automation doing things nobody watched it do, more context than any one person can track on a Tuesday afternoon.
The usual response is to add. Another process, another tool, another coordination layer, another framework with a name. Sometimes that is right. More often the thing that moves the work is five lines on a page that make the next step obvious and the important steps hard to skip.
Handing a team a short list does something else worth having. It tells them the work is knowable, and that somebody thought about which parts matter. A new person gets a way to be useful on their second day. An experienced one gets to stop keeping all of it in their head.
Ninety years after a card saved an airplane nobody could fly, it still does more than most of what we have added since.
Reader discussion