Escalation Decision Tree and Non-Delegable List

What this is

Chapter 5 gives managers a four-question Escalation Decision Tree and a list of six things a manager must never delegate to AI. Chapter 10 adds signs that the Transformation Threshold has been crossed, mandatory human escalation triggers, a tiered escalation protocol, and a ten-box "Signs That This Needs a Human" checklist with a 3+ rule. Together they say when to stop using AI and bring in a person.

Use when / Do not use when

Use whenever a coaching topic is on the edge between skill and something deeper, or when a system needs written escalation rules. Do not use the tree as a scoring instrument. The book's only count rule is the 3+ boxes rule for the Ch.10 checklist, and the book's closing rule for the tree is "When in doubt, escalate to human."

The Instrument

The Escalation Decision Tree (Ch.5, verbatim)

What Managers Must Never Delegate to AI (Ch.5, six items)

1. Career direction conversations. "Should I take this role?" requires human judgment.
2. Performance concerns that could lead to termination. High stakes, legal implications, human required.
3. Trust repair after conflict. When someone feels hurt or betrayed, they need a human to make it right.
4. Ethical dilemmas. When someone is facing a right-versus-right decision, you cannot outsource your judgment.
5. Situations involving mental health concerns. If someone seems depressed, anxious, or struggling, escalate to a human immediately.
6. Anything where the person needs to feel seen by a human. Sometimes people do not need advice. They need to be witnessed.

Signs you've crossed the threshold and need a human (Ch.10, verbatim)

• The conversation keeps coming back to "but who am I if I do this?"
• The person is questioning the goal itself, not how to achieve it.
• Emotions are running high and complex, not just frustration with a skill gap.
• The situation involves other people in complicated ways that require wisdom, not technique.
• Ethics or values are at stake in ways that don't have clean answers.
• The person seems stuck in a way that practice won't fix. Something deeper is blocking them.

If unsure, ask: "Is this about what to do, or about who to be?" / "Is this about execution, or about direction?" / "Is this about skill, or about something deeper?" "If it's the second option, you need a human."

Mandatory human escalation triggers (Ch.10)

• Signs of mental health distress: depression, anxiety, trauma responses
• Career or life decisions with major consequences: job changes, career pivots, relocations
• Complex interpersonal conflict that could escalate: damaged trust, broken relationships
• Ethical dilemmas without clear answers: competing values, loyalty conflicts
• Significant identity transitions: becoming a leader, becoming a parent, recovering from loss

The Escalation Protocol (Ch.10, "When AI coaching should escalate to human")

• IMMEDIATE escalation: mental health concerns (depression, anxiety, trauma); ethical violations or concerns; safety issues (self-harm, harm to others); legal or compliance issues.
• PRIORITY escalation (within 24 hours): identity crisis or major life decision; career direction questioning; complex interpersonal conflict; values misalignment with role; sustained disengagement despite skill development.
• PLANNED escalation (scheduled human time): quarterly career conversations; annual purpose reviews; promotion or role transition support; significant setback processing.

Signs That This Needs a Human (Ch.10, verbatim checklist)

• The person is questioning who they are, not just what they do
• The conversation involves deep emotion (grief, shame, fear)
• Values or ethics are in conflict
• The decision will significantly alter their life direction
• Complex relationships are involved
• The person seems stuck despite skill development
• They're asking "why" more than "how"
• They need to be witnessed, not advised
• You sense something deeper than the presenting issue
• Your gut says "this needs more than AI"

"If you check 3+ boxes, escalate to human."

Why AI fails at Transform work (Ch.10, the book's reasons)

No lived experience; no genuine relationship; no stake in the outcome; cannot witness (one of the deepest human needs is to be seen, "not fixed, not optimized, not coached, witnessed").

Know your limits (Ch.10)

"This is coaching, not therapy. When mental health issues emerge, refer to appropriate professionals." The book also says: "Your job is to support someone's development, not to heal their wounds."

The Trust Imperative (Ch.10)

"AI can surface truth. Humans must make it safe to act on truth. AI can track progress. Humans must witness the person. AI can suggest paths. Humans must honor the choice." "AI can be a tool in service of transformation. AI cannot be the relationship in which transformation happens."

Designing systems that protect the human domain (Ch.10, by level)

• Individual: keep at least one human development relationship; know when to stop asking AI and start asking a person (use the threshold test); do not let efficiency override what you need.
• Manager: protect your 1:1 time; train yourself to recognize threshold moments ("how" shifting to "who"); do not let AI convenience replace your intuition.
• Leader: build human escalation into AI coaching systems; fund human coaching for Transform work; measure what AI cannot measure (trust, purpose, belonging, meaning).

How to run it (Performance Lab suggested practice)

1. Walk the four questions in order for the situation. Write the side for each.
2. Check the six non-delegable items. If the situation is one of them, stop and go to a human.
3. If any of the mandatory triggers applies, treat it as escalate, with the book's IMMEDIATE / PRIORITY / PLANNED timing.
4. If still unsure, complete the ten-box checklist; escalate at 3 or more.
5. Record who the human is, the time frame chosen and what was handed over.

Record sheet (Performance Lab suggested practice)

Situation (no personal or health detail beyond what is needed to act): ______