Why it matters
What this principle is for.
If you only do the things you already know will work, you leave a lot of opportunity on the table. An experiment is not an experiment if you know the outcome. Some of those bets fail. You often learn more from the miss than you would have from a safe success.
Look for whether someone can tell a two-way door from a one-way door. If the decision is reversible, act. More study after you have enough information is waste. If the decision is irreversible, that is when you Dive Deep and slow down on purpose.
Bias for Action also looks like initiative. It is walking down the hall when yesterday's message got silence. It is picking up the next useful task while you wait. It is booking the meeting, showing up on time, and not expecting someone else to do the work you can do.
It also looks like impatience when a customer is hurting. You join the investigation before anyone proves it is your team's bug. You do the right thing for the customer without waiting for a permission slip. Calculated risk is the point. Motion without a target is not.
Calibration
Under, just right, over.
Score whether someone moves with calculated risk and then course-corrects. Under is paralysis or passivity. Over is steamrolling, no focus, and a pile of unfinished work.
| Situation | Under | Just Right | Over |
|---|---|---|---|
| A thread that needs a decision | Lets the thread sit. People keep adding comments instead of making a call. | Names the decision, picks a reversible option with the information already in the thread, and writes what would change it. | Closes the thread with a snap call that ignores a constraint already written there. The team spends the rest of the week undoing the work. |
| Another round of research | Starts more research when the open questions would not change this week's action. | Stops the research once it will not move the decision, starts the reversible work, and leaves a note on what is still unknown. | Skips the afternoon of research that would have answered the one real risk. Someone else inherits the miss. |
| Unblocking yourself | Waits for the owner to return before touching work that is stuck. The week goes idle. | Tries a cheap path that can be reversed, and tells the owner what they did. | Forces a path the owner had already ruled out. The owner spends their first day back cleaning it up. |
| The plan breaks midweek | Freezes until a new plan is approved. The remaining days are spent waiting. | Picks a next step that keeps learning cheap, and updates the people who were on the old path. | Scraps the plan and starts a new one the same day without checking who is already midstream. Their work no longer connects. |
| One more review before acting | Adds another review because it feels safer. The action waits for a slot that is not this week. | Takes the review that would catch a hard to reverse miss, and skips the rest so the work can start. | Skips the review that was going to catch a known risk. The people who would have flagged it find out after the cost is real. |
| A choice that can be undone | Treats a reversible choice like a strategy decision. Days go to alignment the choice does not need. | Picks one, starts, and watches. Changes it if the first days show it is wrong. | Flips the choice twice in one week. People stop trusting the direction and redo work each time. |
| Work that is ready enough to share | Holds the draft until it is polished. Reviewers have nothing to react to this week. | Shares the rough version early enough that feedback can still change it, and says what is still unfinished. | Sends half formed work to a wide list so it looks like progress. People spend time commenting on parts they already knew were wrong. |
| A small fix in front of you | Files it for later and lives with the friction for the rest of the week. | Fixes the reversible part now, and leaves a note if a larger change still needs a decision. | Reworks the surrounding process in the same sitting. People who used the old path are surprised and lose time. |
Examples
What it looks like in the work.
Two-way door, then correct
A team can A/B a change, ship it to a slice of customers, and roll it back in an afternoon. The leader does not commission a six-week study. They ship the slice, read the result, and either widen it or revert. That is calculated risk. The irreversible launch to every customer is the door you do not kick open.
The experiment that taught more than a win
A team spends three years on a new front end for a capability they licensed. Nobody uses it. They shut it down. The useful story is not the failure. It is that they were willing to run a real experiment, call the miss, and stop. If you only fund the sure things, you do not invent.
Do not wait for the alert
Metrics start to look wrong in the afternoon. The person who has been watching them joins the call before the alert fires, and starts the investigation before anyone assigns a team. Bias for Action is that impatience when a customer is about to get hurt, not a personality that cannot sit still in a meeting.
In the role
Individual and manager.
Individual
An individual with Bias for Action unblocks themselves. They send the note, then they walk over. They keep a next useful task ready for the idle time. They can tell you which recent decision was reversible, what they did with incomplete data, and what they changed when the result came back.
Manager
A manager with Bias for Action removes the approval theater that turns two-way doors into projects. They expect calculated risk, they do not punish a well-run miss, and they stop the person who is generating motion without finishing. They also know when to force a pause because the door only swings one way.
Go deeper
Questions that make the principle concrete.
- Was the last decision you delayed a two-way door, and if so, why did you wait?
- What is the smallest reversible step you could take this week with the information you already have?
- When a customer was hurting, did you join before the work was assigned to you, or did you wait for proof it was your bug?
- Whose input do you actually need before you move, and whose input are you collecting only to avoid being wrong?
- What did you finish in the last month, and what did you start and abandon?
- How do you tell a calculated risk from doing the work at any cost?
- What did the last failed experiment teach you that a safe success would not have?
- When yesterday's message got silence, what did you do instead of sending a second one into the same void?
- Where are you still studying after you already have enough information to act?
From the blog
Writing that goes deeper.
- One-Way and Two-Way Doors
If you can walk it back, act. If you cannot, that is when you slow down.
Related