**We give most decisions more stress than they actually deserve.**
A small number of decisions justify days or weeks of serious deliberation.
But most decisions don't.
The problem is that we frequently treat both types of decisions with the same decision-making process. We deliberate over a major strategic commitment and a small operational experiment as though both carry the same consequences.
They don't.
Better decision making begins by asking one question:
**How reversible is this decision?**
A reversible decision can usually be changed, tested, adjusted, or undone without significant cost. An irreversible decision creates consequences that may be expensive or impossible to reverse.
That difference should shape the amount of analysis required, the speed of the decision, and—most importantly—**where the authority to make it should sit.**
And that is the point where a decision-making problem becomes a question of decision architecture.
## What Makes a Decision Reversible or Irreversible?
A **reversible decision** gives you the ability to change course at relatively low cost. It allows you to make a choice, gather feedback, and change direction if necessary.
Examples of reversible decisions include:
* Experimenting with a new meeting structure
* Testing a different workflow
* Testing a small marketing initiative
* Testing a new process with one team
* Testing the delegation of a responsibility
* Running a limited pricing test
An **irreversible decision** creates consequences that are difficult or costly to undo.
Examples include selling a business, entering a major long-term agreement, committing substantial capital, or taking an action that could permanently harm trust or reputation.
The rule is simple:
**Hard-to-reverse decisions deserve greater scrutiny. Easy-to-reverse decisions should move closer to the people who have the best information.**
This is the foundation of what I call the **Decision Reversibility Framework.**
## Why Are Reversible Decisions So Easy to Overthink?
Reversible decisions rarely feel reversible while we're making them.
A manager considering a process change may focus on everything that could go wrong.
A leader considers delegating authority and worries about losing control.
An entrepreneur delays an experiment while waiting for additional evidence.
Instead of acting, uncertainty creates a demand for more information.
Another meeting.
One more spreadsheet.
Another round of feedback.
Another person who needs to sign off.
Eventually, the decision-making process costs more than making the wrong reversible decision would have cost.
This produces **decision latency**—the delay between having sufficient information and taking action.
One delayed decision may not matter. Hundreds of them do.
If every reversible decision must travel up the hierarchy, senior leadership becomes the constraint.
This does not necessarily make the organization safer.
It has simply centralized hesitation.
## The Decision Reversibility Framework
Before deciding what action to take, evaluate the decision through five questions.
### 1. How Reversible Is the Decision?
Don't treat reversibility as a yes-or-no question.
Ask what reversal would cost.
A new meeting format can be reversed quickly. A company-wide technology implementation may technically be reversible but financially and operationally painful to undo.
Think of reversibility as a spectrum rather than a binary category.
The greater the cost of reversing course, the more scrutiny the original decision deserves.
### 2. What Happens If You're Wrong?
Distinguish actual consequences from temporary discomfort.
Failure in a small experiment may mean a few weeks of lost effort.
A major strategic mistake may expose the organization to substantial financial, legal, operational, or reputational risk.
Ask:
**If we're wrong, what do we really lose?**
The size of the downside should influence how much uncertainty you're willing to accept.
Low downside allows faster action.
When the downside is significant, greater scrutiny is appropriate.
### 3. How quickly will reality give us feedback?
For some decisions, action produces useful information faster than additional analysis.
A sales-script experiment can start producing feedback in a matter of days.
You can test a meeting format for several weeks and see whether productivity improves.
When feedback is available quickly, you can move away from:
**Analyze → Analyze → Analyze → Decide**
to:
**Decide → Test → Learn → Adjust**
For reversible decisions, action isn't merely execution.
**Action itself produces evidence.**
### 4. Can We Limit the Downside?
A new decision doesn't need to be rolled out everywhere at the same time.
Pilot the workflow with one team.
Offer the product to a small customer segment.
Pilot the policy for one month.
Delegate authority within clear boundaries.
Instead of demanding certainty, contain the downside.
**Before responding to uncertainty with more analysis, see whether you can reduce the size of the commitment.**
This can turn seemingly high-risk decisions into manageable experiments.
### 5. Where Should Decision Authority Sit?
This is the question most decision-making frameworks miss.
If the decision can be reversed cheaply, produces fast feedback, and has limited downside, why should senior leadership need to approve it?
Often, it doesn't.
Authority should move closer to the people who understand the situation best.
The leadership question then shifts from:
**"What should I decide?"**
and replaces it with:
**"Why does this decision require me?"**
At that point, decision making becomes an issue of power architecture.
## A Decision Architecture Matrix
Use reversibility and cost of error to determine the appropriate decision process.
| Decision Type | Reversibility | Cost of Error | Best Response | | |
| ---------------------- | ------------- | ------------- | -------------------------------- | ----------------------------------------------------- | ------------------------------------------- |
| more info **Experiment** | High | Low | Decide quickly and test |
| **Delegated Decision** | High | Moderate | Delegate within clear guardrails |
| **Strategic Bet** | Low | Moderate–High | Increase analysis before committing |
| **Critical Decision** | Very low | High | Increase scrutiny and deliberate carefully |
The common mistake is applying **Critical Decision behavior** to almost everything.
Increasing analysis doesn't guarantee higher-quality decisions.
Sometimes the only result is a slower organization.
## How Much Information Do You Need Before Deciding?
If a decision can be reversed and failure is manageable, waiting for perfect information is usually unnecessary.
A useful principle is to act once you have enough information to make a reasonably informed choice and enough protection to survive being wrong.
Jeff Bezos popularized the idea of reversible "two-way door" decisions and has argued that many decisions can be made with around 70% of the information you would ideally want.
The deeper principle is this:
**Reversible decisions need guardrails more than they need certainty.**
When testing is inexpensive, feedback is fast, and reversal is manageable, action may produce more value than additional analysis.
Experimentation becomes part of the decision-making process.
## Use Guardrails Instead of More Approvals
The leadership implications become clear at this point.
Suppose employees routinely bring decisions to you before taking action.
You may think:
**They need to become more decisive.**
Yet asking for permission may be the rational response.
If decision rights are unclear, asking for approval becomes the safest behavior.
The answer isn't another speech about empowerment.
Fix the structure.
Instead of controlling each decision individually, establish clear boundaries of authority:
"You have authority to resolve customer issues below this financial limit."
"Process changes can be tested within your team for a 30-day period."
"Experiments don't require additional approval unless they affect legal, security, or brand standards."
People now know where their authority begins and ends.
People don't need permission for every action. They need judgment within an established territory.
**An approval system controls individual decisions.**
**Architecture determines how decisions are made repeatedly.**
Only the second approach scales.
## A Two-Minute Test for Better Decisions
Before your next significant decision, spend two minutes classifying it.
Ask:
**1. Is it reversible?**
Could we change direction without significant damage?
**2. What would failure cost?**
Define the actual downside.
**3. How quickly will we know?**
Identify how quickly useful feedback will arrive.
**4. Can we limit the first commitment?**
Reduce, pilot, stage, or limit the initial commitment.
**5. Who should own the decision?**
Move decision authority toward the best available information.
Your answers should determine what happens next:
**Easy to reverse + low cost of error → Decide.**
**Easy to reverse + uncertain result → Test.**
**Easy to reverse + someone else has better information → Delegate.**
**Hard to reverse + high downside → Deliberate.**
Many difficult decisions become simpler once you classify them correctly.
## Decision Speed Is a Power Advantage
The objective isn't speed at all costs.
That would create recklessness.
The objective is to give every decision the level of scrutiny its downside deserves.
Some decisions justify extensive analysis, opposing viewpoints, scenario planning, and scrutiny.
Other decisions need little more than a controlled test.
Beyond decision speed, there is a more important leadership implication.
If every important decision comes to you for approval, it may feel like evidence of your authority.
Structurally, the opposite may be true.
The system has become dependent on your intervention.
Real organizational power isn't measured by the number of decisions that require a leader's approval.
Real organizational power appears when the system can produce good decisions **without requiring continuous intervention from the person at the top.**
Reversibility therefore matters far beyond individual decision making.
It shapes where decision-making authority should be located.
It determines how efficiently knowledge becomes action.
And it determines whether an organization operates through judgment or permission.
So before asking:
**"What should we decide?"**
start with:
**"How reversible is the decision, and who is best positioned to make it?"**
Most decisions don't need more anxiety or deliberation.
What they need is better decision architecture.
In *The Architecture of POWER*, I explore how decision rights, authority, incentives, information, and invisible systems shape the way power actually works inside organizations.
**Real power isn't just about who decides. It's built into the architecture through which decisions get made.**