DESC & the Ladder of Inference
DESC and the Ladder of Inference are two smaller but high-leverage tools for difficult conversations. DESC is a compact script for delivering assertive messages in the heat of the moment. The Ladder of Inference is a diagnostic model for understanding how disagreements often trace back to different data, different interpretations, or different conclusions — rather than to bad faith.
Together they cover the delivery (DESC) and the thinking (Ladder) sides of staying productive in a difficult conversation.

DESC
Origin
DESC was developed by Sharon and Gordon Bower in their 1976 book Asserting Yourself, which brought assertiveness training out of the clinical-psychology world and into general workplace use. It was designed for situations where the communicator has been non-assertive (silent, hinting, passive) for too long and needs a structured way to make an assertive move without escalating into aggression. DESC is now standard in nursing, healthcare, and crew-resource-management (CRM) training, where crisp assertive communication across hierarchical gradients is literally a safety issue. It has migrated into tech via on-call and incident training.
The Framework
D - Describe: The specific situation, factually
E - Express: Your feelings, reaction, or concern about it
S - Specify: What you want to happen differently
C - Consequences: What will change (positive) if they agree;
what follows if they don't
The power of DESC is its compactness. A well-delivered DESC is 4-6 sentences and produces a crisp, assertive move:
D: "In the last three on-call handoffs, I've received the pager without
a written summary of active incidents."
E: "I've felt scrambled during the first hour each time, and I'm
worried we'll miss something critical."
S: "Going forward, could we agree to a written handoff in the
#oncall-handoff channel, using the template, before the shift
change?"
C: "If we do, I think it'll reduce MTTA for both of us. If it keeps
not happening, I'll raise it with the tech lead as a process
gap, which I'd rather not do."
How to Use It
1. Prepare the four pieces in writing before the conversation.
2. Deliver in order; do not rearrange (the sequence matters).
3. Use "I" language for Express ("I felt", not "You made me feel").
4. Specify a concrete, doable change, not a personality overhaul.
5. Be honest about Consequences — do not threaten things you won't do.
When DESC Works Best
- Repeat-offender situations where softer attempts have failed
- Peer-to-peer conversations where you need to be more assertive
- On-call, incident, or safety-critical contexts
- Situations where you have a clear ask with a concrete alternative
When DESC Does Not Work
- Complex, multi-issue conversations (too compact for nuance)
- Highly emotional conversations where Crucial Conversations' fuller toolkit is needed
- Situations where you do not actually have leverage for the Consequences step
- When the other person has not had a chance to share their side; DESC is a delivery tool, not a dialogue tool
Common DESC Failure Modes
- Describe becomes attack — "You never do the handoff" is not Describe; it is accusation. Describe must be factual.
- Express becomes blame — "You make me feel anxious" is blame. "I feel anxious about missing incidents" is Express.
- Specify becomes demand — "Do the handoff" is a demand. "Could we agree to do the handoff using the template" is a proposal.
- Consequences become threats — "Or I'll report you" is a threat; it creates adversarial dynamics. Consequences should be honest, proportional, and as cooperative as possible.
- DESC without listening — Delivering DESC without pausing to hear the response makes it a monologue.
The Ladder of Inference
Origin
The Ladder of Inference was developed by Chris Argyris, a Harvard organizational psychologist who spent his career studying how smart people make themselves dumb in organizational settings. Peter Senge popularized it in The Fifth Discipline (1990) as one of the core tools of his systems-thinking and learning-organization practice. It is now a standard facilitation tool in leadership development, executive coaching, and organizational change work.
The Framework
The Ladder describes the (often invisible) mental process by which we move from data to action:
[Action] <-- 7. I take action
^
|
[Beliefs] <-- 6. I adopt beliefs
|
[Conclusions] <-- 5. I draw conclusions
|
[Assumptions] <-- 4. I make assumptions
|
[Meaning] <-- 3. I add meaning
|
[Selected Data] <-- 2. I select data
|
[Observable "Facts" & Experiences] <-- 1. Available data
Each rung is a step further from the shared observable reality. By the time we reach Actions and Beliefs, two people can have radically different conclusions from the same starting data.
Why It Matters
Most difficult conversations fail not because people disagree about facts, but because they are each standing at the top of their own ladder, arguing about Beliefs or Actions, with no shared visibility into how they got there. Climbing back down the ladder — and inviting the other person to do the same — often resolves the conflict without anyone having to "win."
Example: The Ladder in Action
Two engineers disagree about a teammate:
Engineer A's ladder:
1. Observable data: Sam did not respond to my Slack message on Friday.
2. Selected data: Sam has not responded to 3 of my messages this
month.
3. Meaning added: Sam is not engaging with my work.
4. Assumption: Sam does not value my contributions.
5. Conclusion: Sam is blocking my growth.
6. Belief: Sam is a bad manager.
7. Action: Start looking for another job.
Engineer A is now convinced Sam is a bad manager. Meanwhile, Sam's
ladder might be:
1. Observable data: Received Engineer A's Slack at 6pm Friday,
during my kid's birthday.
2. Selected data: Have been overwhelmed; slow on non-urgent
things.
3. Meaning added: Non-urgent messages can wait until Monday.
4. Assumption: Engineer A would ping again if urgent.
5. Conclusion: No immediate response needed.
6. Belief: Setting boundaries around family time is
healthy.
7. Action: Will respond Monday.
Both ladders start from the same fact. The conflict lives in the
rungs between.
How to Use It
The Ladder is used in two directions:
Climbing Down (for yourself):
When you notice you've formed a strong judgment about someone:
1. What am I doing / planning? (Action)
2. What do I believe about them? (Belief)
3. What conclusion did I draw? (Conclusion)
4. What did I assume? (Assumption)
5. What meaning did I add? (Meaning)
6. Which data did I actually select? (Selected data)
7. What other data was available that I ignored? (Observable)
Inviting Them to Climb Down (with another person):
Use phrases that make ladder-climbing explicit:
"The story I'm telling myself is..."
"Help me understand the data you're working from..."
"I might be making an assumption here — can I check it?"
"What am I missing?"
Tech & Company Example
A product manager and an engineer are in conflict over scope. The PM believes the engineer is "always pushing back"; the engineer believes the PM is "constantly changing direction." Both are at the top of their ladders.
Facilitated climb-down:
Moderator: "Let's each share the actual data this week. PM first."
PM: "Specifically: on Tuesday you pushed back on the settings-
page ticket. On Wednesday you pushed back on the analytics
integration. That's what I'm working from."
Engineer: "Tuesday I pushed back because the ticket was tagged 'small'
but actually needed a data migration. Wednesday I asked
questions about the analytics because the spec changed
between Monday and Tuesday and I wasn't sure which version
was final."
Moderator: "So the underlying data is: PM saw two pushbacks and
concluded 'always'. Engineer saw two spec issues and
concluded 'constantly changing'. The 'always' and
'constantly' are conclusions, not data."
Once both ladders are visible, the conversation moves from "who is being difficult" to "how do we handle scope changes in specs." A solvable problem.
When It Works
- Any conflict that has hardened into mutual character judgments
- Facilitated conversations (the ladder is a great shared visual)
- Leadership coaching when the leader has developed strong negative views
- Retrospective discussions about past incidents ("what did we each conclude?")
When It Does Not Work
- When the underlying data is genuinely different (the issue is data access, not interpretation)
- When one party is not willing to examine their own ladder
- In fast-moving operational decisions (too heavy-weight)
- When the ladder becomes a conversation-stopping gotcha ("you're climbing your ladder!") rather than a shared tool
Common Ladder Failure Modes
- Weaponized Ladder — Using the framework to dismiss someone's conclusion ("you're just selecting data") without examining your own. The Ladder only works if all parties use it on themselves first.
- Ladder theatre — Performing the framework without actually examining beliefs. "I acknowledge I climbed my ladder" as a rhetorical move.
- Reducing everything to interpretation — Sometimes conclusions are correct. The Ladder's purpose is not to dissolve all conclusions into relativism; it is to surface the path so we can test it.
- Ignoring the action — Climbing down the ladder and concluding "it's all just perception" without still deciding what to do differently.
How DESC and the Ladder Fit Together
1. Notice you have a strong reaction to someone's behavior.
2. Use the Ladder: climb down; check your assumptions and conclusions.
3. If, after climbing down, you still believe an assertive message is
warranted: use DESC to deliver it.
4. Invite the other person to share their ladder before reacting
to yours.
The Ladder is diagnostic (should I say something?); DESC is delivery (how do I say it?).
Further Reading
- Sharon Bower & Gordon Bower — Asserting Yourself (DESC's canonical source)
- Chris Argyris — Overcoming Organizational Defenses (Ladder in its original organizational-learning context)
- Peter Senge — The Fifth Discipline (popularized the Ladder)
- Rick Ross in The Fifth Discipline Fieldbook — detailed applied Ladder exercises
- Douglas Stone, Bruce Patton, Sheila Heen — Difficult Conversations (uses related concepts under different names)