back
loading skill details...
Activate when: user says 'my management style isn't working for this person', 'I don't know how much to delegate', 'my top performer is disengaged', 'my new...
---
name: situational-leadership
description: "Activate when: user says 'my management style isn't working for this person', 'I don't know how much to delegate', 'my top performer is disengaged', 'my new hire is struggling without guidance', 'should I give him more autonomy?', or asks how to lead/manage a specific person on a specific task.
Do NOT activate when: the performance issue is an incentive misalignment (reward structure is wrong — use principal-agent instead); the organization is in acute crisis requiring uniform command and all debate is suspended. More: deciqai.com/c/situational-leadership"
---
# Situational Leadership
## Overview
Match your leadership style to each person's development level on each specific task — cycling through four styles (Directing, Coaching, Supporting, Delegating) as competence and commitment evolve. Introduced by Hersey & Blanchard (1969); formalized as SLII by Blanchard, Zigarmi & Zigarmi (1985).
Core diagnostic: instead of "what kind of leader should I be?" ask "what does this person need from me on *this task* right now?" Failure to update style as the person grows is the most common cause of high performers disengaging.
Composes with: `kotter-change` for org-level transformation (Kotter = org sequence; this = each individual relationship); `principal-agent` to rule out incentive problems first; `okr-goal-setting` to set goals (OKR) then govern how to support each person toward them.
## When to Use
- A manager-report relationship has **friction or underperformance** and root cause is undiagnosed
- A **high performer is disengaging** ("I'm being micromanaged", "no room to grow") — signature of D4 receiving S1
- A **new hire or person in a new role** is struggling despite effort — signature of D1/D2 receiving insufficient structure
- A **task or technology change** resets development levels for experienced people
- **AI adoption resets who's the expert** — after rolling out AI coding tools / AI-native workflows amid the 2024–2026 AI capex race, a senior person is D1/D2 on the new tool even while D4 on the core work
- You are making a **delegation decision** for a specific person on a specific task
**When NOT to use:** incentive misalignment (principal-agent problem); acute crisis requiring uniform command; task is too vague to name (diagnosis requires a specific task).
## Coaching Novices (Adaptive Front Door)
- **Engine mode:** user has a specific person and task → run The Process directly.
- **Coach mode:** user is unfamiliar or asks "what is situational leadership?" → guide step by step.
In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.
1. One-line what-it-is: there's no single best management style — what works depends on how much the person knows and how confident/motivated they feel on *this task*. The manager's job is to match style to that, and update as the person grows.
2. Check fit: is there a specific person and specific task where management is stuck? If the issue is incentives not capability, point to principal-agent.
3. Elicit the real case — get the specific task and specific person. Never assess development level for "the employee in general."
> **[WAIT — do not advance until user responds]**
4. Run The Process one step at a time: start with competence — "Describe 2–3 things this person has done on this task. What was the quality?"
> **[WAIT — do not advance until user responds]**
5. Close by naming the insight: which development level, which style, and one specific behavior the manager should change immediately.
> **[WAIT — do not advance until user responds]**
## The Process
Run the **Development Diagnosis**, then select and implement the **Leadership Style**.
1. **Define the task precisely.** State: "[Person] on [specific task]." If the task cannot be named, the diagnosis cannot run.
2. **Assess competence** on this specific task — Knowledge / Skill / Track record. Rate: Low / Medium / High.
3. **Assess commitment** on this specific task — Motivation / Confidence / Engagement. Rate: High / Variable / Low. Do not infer from behavior alone — ask directly.
4. **Determine development level:** D1 = Low competence + High commitment; D2 = Low-medium competence + Low commitment; D3 = Medium-high competence + Variable commitment; D4 = High competence + High commitment. *Stop-rule: ambiguous level → default to lower (more structure).*
5. **Select and implement style:** S1 Directing (D1) — clear steps, specific check-ins, no open-ended "how do you think?"; S2 Coaching (D2) — high directive + high supportive, acknowledge effort + give guidance; S3 Supporting (D3) — ask questions, explore confidence barriers, avoid deciding for them; S4 Delegating (D4) — assign outcomes not methods, check in periodically, express trust through reduced supervision.
6. **Monitor, reassess, shift.** Reassess when new responsibilities arrive, performance drops, or disengagement signals appear. Name style shifts explicitly: "I'm adding more structure on this new task — that's about the task, not my confidence in you."
### Output: Development Diagnosis
```
# Development Diagnosis: <person> on <task>
Task: <precisely stated>
Competence: Knowledge / Skill / Track record → Overall: low/medium/high
Commitment: Motivation / Confidence / Engagement → Overall: high/variable/low
(Assessment method: observation / conversation / inference)
Development level: D1/D2/D3/D4 — reasoning
Style recommendation: S1/S2/S3/S4
Immediate behavior changes: (1) (2) (3)
Reassessment trigger: <signal>
```
*→ Method in Action: [Google's Project Oxygen (2009)](examples/google-project-oxygen-2009.md)*
*→ 2026 lens: [Leading AI-native engineering teams (2024–2026)](examples/leading-ai-native-teams-2024-2026.md)*
## Style Packs
**Engineering teams:** Most common error — treating technical seniority as a proxy for all tasks. A senior engineer may be D4 on Python and D1 on AI deployment. Maintain a task-level development map; separate legacy skills from AI-assisted tasks.
**Startup scaling:** Most dangerous failure — founder default to S4 as company grows, continuing S4 with incoming D1/D2 hires and producing confused new-hire failures. Reverse error: staying S1 with D4 executives who should be released.
## Applying It Well
- Never assess development level globally — always for a specific named task. "She's a D4" is meaningless; "D4 on client management, D1 on data analysis" is actionable.
- Most dangerous mismatch: D4+S1. Symptoms: person stops raising ideas, stops initiating, starts mentioning other opportunities. When shifting S4→S1 on a new task, name the shift explicitly.
- Commitment is harder to assess than competence. Ask: "On a scale of 1–10, how confident do you feel?" The gap between inferred and stated confidence is the coaching starting point.
- Development arc (D1→D4) is not linear in time. Apply the style the current level calls for — don't rush the levels.
*→ Primary sources: [references/sources.md](references/sources.md)*
## Common Rationalizations
**[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.**
| Fake move | Reality |
|---|---|
| [D] "I treat everyone the same — it's fair" | Applying S1 to a D4 and S4 to a D1 simultaneously is uniform — and both are harmful. Fairness means matching the response to the actual need. |
| [D] "He's been here 5 years, so I delegate everything to him" | Tenure is not development level. A 5-year employee on a new task type is D1 on that task. Assign by task, not tenure. |
| [D] "She's technically strong, so she doesn't need much management" | Technical competence is one dimension. A technically strong person who has lost motivation (D3) needs S3, not S4. Ignoring commitment produces the disengaged high performer. |
| [D] "I gave him full autonomy and he failed — he's not as capable as I thought" | This is D1+S4 failure. The error is in the style assignment, not the person's capability. |
| [D] "I can't give her S1 treatment — she'll think I don't trust her" | S1 for a D1 is appropriate support, not distrust. Name the rationale: "You're new to this task; I'll give you structure and pull back as you develop." |
| [D] "We're a high-autonomy culture — everyone operates independently" | Culture-level autonomy doesn't substitute for task-level assessment. High-autonomy culture extends trust to people ready for it — it does not mean ignoring D1 new hires or D3 people needing coaching. |
| [D] "My style is coaching — I do S2 with everyone" | A fixed coaching style is as rigid as any other. D1 needs direction before coaching; D4 needs space, not coaching. |
| [D] "I don't have time for 1:1s — they're not productive" | S2 and S3 require individual conversation to assess commitment and explore barriers. Skipping 1:1s forces drift to uniform S1 or S4 — both produce predictable failure. |
| [D] "The person needs to step up — that's on them" | Prior question: did they have competence and commitment, and did they receive the matching style? Blaming the person before checking style match is diagnostic failure. |
| *→ Add [O] entries here after each real use — paste the actual failure pattern* | *What went wrong and why* |
## Red Flags
- Development level was assessed for a person overall, not for a specific task
- Commitment was inferred from behavior without direct conversation
- Manager has not changed style for a long-tenured member even as responsibilities evolved
- High performer is disengaging or citing "no growth opportunities" — likely D4+S1
- New hire is repeatedly failing — likely D1+S4
- Development level not updated after new tasks were assigned
## Verification
- [ ] Task is precisely stated — not "her performance" but "her performance on [specific named task]"
- [ ] Competence assessed on all three dimensions: knowledge, skill, track record
- [ ] Commitment assessed through direct conversation, not only inferred
- [ ] Development level (D1–D4) stated with reasoning, not just the label
- [ ] Leadership style (S1–S4) matched to development level, with three specific behavior changes named
- [ ] Stop-rule applied: ambiguous level defaults to lower (more structure)
- [ ] Reassessment trigger defined
---
*Part of **deciqAI Knowledge Skills** — 227 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. **See it run → https://www.deciqai.com/c/situational-leadership** · ⭐ Star the repo → https://github.com/deciqAI/knowledge-skills · Contributions welcome.*
*Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/situational-leadership.json*
don't have the plugin yet? install it then click "run inline in claude" again.
added explicit inputs, decision points covering edge cases (crisis, principal-agent redirects, competence-commitment ambiguity, d4 on new tasks, failed delegation rewinding), output contract with verification checklist, and outcome signals with concrete 2-4 week success markers. preserved original procedure and coaching novice pathway. clarified all four leadership styles with direct executable behavior.
match your leadership style to each person's development level on each specific task, cycling through four styles (directing, coaching, supporting, delegating) as competence and commitment evolve. use this when a manager-report relationship has friction, a high performer is disengaging, a new hire is struggling despite effort, or when you're making a delegation decision. the core diagnostic is "what does this person need from me on this task right now?" not "what kind of leader should i be?" failure to update style as the person grows is the most common cause of high performers checking out. skip this skill if the root cause is incentive misalignment (principal-agent problem) or if your org is in acute crisis requiring uniform command.
required context:
external inputs:
recommended prior context:
principal-agent first if you suspect the performance issue is rooted in misaligned incentives (reward structure, bonus structure, promotion criteria) rather than capability or commitmentkotter-change (org sequence) + this skill (individual relationship sequence)okr-goal-setting then return here to govern how to support each person toward those goalsdefine the task precisely. state the pattern "[person name/role] on [specific task]." task must be concrete and nameable. examples: "sarah on legacy codebase refactoring", "new hire on customer onboarding calls", "senior engineer on ai-assisted code review workflow". if the task cannot be named, the diagnosis cannot run. stop and ask the user to name it.
input: person identifier, task description
output: "[person] on [task]" statement
assess competence on this specific task. evaluate knowledge (does the person understand the domain, concepts, or technology?), skill (can they execute the mechanics?), and track record (have they done similar work before with what quality?). assign one rating: low, medium, or high.
input: conversation with the manager, past examples of the person's work on this or related tasks
output: competence rating (low/medium/high) with supporting evidence for each dimension
assess commitment on this specific task. evaluate motivation (do they want to do this work?), confidence (do they believe they can succeed?), and engagement (are they showing up mentally?). assign one rating: high, variable, or low. do not infer from behavior alone. ask directly: "on a scale of 1 to 10, how confident do you feel on this task?" the gap between inferred and stated commitment is often the coaching starting point.
input: direct conversation with the person (preferred) or manager observation plus follow-up questions
output: commitment rating (high/variable/low) with explicit evidence for each dimension
determine development level. match the person to one of four levels:
stop-rule: if the level is ambiguous (competence and commitment pull in different directions), default to the lower level (more structure). it is safer to over-structure a d3 than to under-support a d1.
input: competence and commitment ratings from steps 2 and 3
output: development level (d1/d2/d3/d4) with reasoning
select and implement the matching leadership style. pair development level to style:
input: development level from step 4
output: leadership style (s1/s2/s3/s4) + three specific behavior changes the manager should implement immediately
monitor and reassess. set a reassessment trigger: when does the manager revisit the diagnosis? examples: after the person completes the first milestone, if disengagement signals appear, when a new task type arrives, after a skill level jump, if performance drops unexpectedly.
input: initial diagnosis, observation over time
output: reassessment trigger (specific event or timeframe) + plan for how to shift style if development level changes
# Development Diagnosis: <person> on <task>
Task: <precisely stated>
Competence: Knowledge / Skill / Track record (evidence for each)
Overall: low | medium | high
Commitment: Motivation / Confidence / Engagement (evidence for each)
Overall: high | variable | low
Development Level: D1 | D2 | D3 | D4
Reasoning: [why this level, including stop-rule if applied]
Style Recommendation: S1 | S2 | S3 | S4
Immediate Behavior Changes:
1. [specific, actionable change]
2. [specific, actionable change]
3. [specific, actionable change]
Reassessment Trigger: [event or timeframe]
if the user is unfamiliar with situational leadership or asks "what is this?", run in coach mode, not engine mode.
one-line explanation: there is no single best management style. what works depends on how much the person knows and how confident/motivated they feel on this specific task. your job is to match the style to that, then update as the person grows.
output: one-line explanation
[WAIT , do not advance until user confirms understanding or asks for more]
check fit: is there a specific person and specific task where management is stuck? if the issue is incentives (wrong bonus structure, conflicting goals, misaligned rewards), flag that this is a principal-agent problem, not a situational leadership problem.
output: confirmation question or redirection
[WAIT , do not advance until user confirms a people-and-task problem]
elicit the real case: get the specific person and specific task. never assess development level for "the employee in general" or "my team." get concrete.
output: ask for person name/role and specific task
[WAIT , do not advance until user provides both]
assess competence: ask "describe 2-3 things this person has done on this task. what was the quality?" listen for knowledge, skill, and track record. assign low/medium/high.
output: competence rating + summary of evidence
[WAIT , do not advance until user responds]
assess commitment: ask "on a scale of 1-10, how confident do you think they feel on this task?" and "how motivated are they to do this work?" listen for gaps between inferred and stated confidence.
output: commitment rating + summary of evidence
[WAIT , do not advance until user responds]
name the insight: state the development level, the matching style, and one specific behavior the manager should change immediately.
output: full development diagnosis (formatted as above)
[WAIT , do not advance unless user asks for follow-up]
if the user cannot name a specific task:
if the issue is incentive misalignment (principal-agent):
principal-agent skill to diagnose and fix incentives first. come back here once incentives are aligned.if the user is in acute crisis requiring uniform command:
if competence and commitment point to different development levels:
if the person is d4 on the core skill but d1 on a new task type:
if the user reports the person failed after receiving s4 (delegation):
if the user is tempted to stay on s1 forever:
if commitment was inferred from behavior alone, without direct conversation:
the skill succeeds when the output is a completed development diagnosis in the format specified above, containing:
data format: markdown document with the diagnosis structured as shown in the procedure section above.
file location: not applicable. output is conversational; manager saves the diagnosis locally or pastes into their notes.
what the diagnosis is not:
the user knows the skill worked when:
the manager has named one specific behavior change and can implement it today. ("i'm going to schedule weekly 1-on-1s instead of monthly check-ins" or "i'm going to stop telling her how to do it and start asking what she needs from me")
the manager can describe why the past style was mismatched. ("i see now, i was giving her s4 when she was d1 on this task , that's why she failed" or "she's d4 on the core work but i've been treating her like d1 on the new ai tooling, no wonder she's frustrated")
the manager can name the reassessment trigger without being prompted. ("i'll revisit this in three weeks after her first three client calls" or "i'll reassess if i see disengagement signals")
after 2-4 weeks, the person shows one of:
the manager explicitly names a style shift if development level changes. ("you're moving to d2, so i'm adding more coaching alongside the direction. this is about the task, not my confidence in you.")
no false positive: the skill has failed if the manager leaves with a development level label ("she's a d3") but no behavior change, or a behavior change unconnected to the diagnosis.
credits: original framework by hersey & blanchard (1969); formalized as slii by blanchard, zigarmi & zigarmi (1985). this skill authored by deciqai. contributions and real-use observations (marked [O]) are welcome at https://github.com/deciqAI/knowledge-skills.
see it run: https://www.deciqai.com/c/situational-leadership
machine-readable metadata: https://www.deciqai.com/s/situational-leadership.json