Teaching Case Development Skill

This workflow applies to both primary cases (interview-based) and secondary cases (public information-based). The only difference is in the information collection stage.

Trigger Words and Standard Actions

The user does not need to use exact trigger phrases. Identify the current stage from conversation context and execute the corresponding actions. Trigger words are shortcuts, not the only entry point.

"New case: [company], [course direction], primary/secondary"

Stage 0–1. This is the most critical stage — it determines the ceiling of the case's teaching value. Do not rush.

Execute three parallel research lines, then synthesize:

Line A — Company deep scan: founder background, milestones, financials, core tech/products, business model, customer structure, key strategic decisions, org structure, recent events. Search in both Chinese and English (prospectus, annual reports, analyst reports, media).

Line B — Industry panorama: industry definition, tech routes, global/China market size and growth, supply chain structure, competitive landscape (detailed competitor profiles), downstream market analysis, barriers, drivers, major events.

Line C — Topic novelty assessment: Search Harvard Business Publishing, Case Centre, CEIBS/Tsinghua/Peking/CKGSB case libraries for: (1) existing cases on this company and their angles, (2) same-industry cases, (3) same-topic cases (e.g., "digital transformation", "going global"). Identify 5–7 potential case directions without pre-constraining to a single course.

Before evaluating, classify each potential direction by case type (Ellet's taxonomy):

• Decision case — protagonist must make a choice (most common, default)
• Problem diagnosis case — root cause must be identified
• Evaluation case — a past decision/strategy must be assessed

This classification shapes everything downstream: question design, information needs, and narrative structure. Most cases are decision cases, but knowing the type early prevents structural confusion later.

Then evaluate each direction using the 5-dimension matrix. See references/topic-evaluation.md for the full matrix (novelty, teaching value, theoretical richness, information availability, discussion scope).

Output:

1. Background knowledge synthesis (edited and analyzed, not raw dump), with information gap list
2. Topic evaluation report with ratings, recommended ranking, integration proposal, and core decision point design
3. For primary cases: interview outlines per recommended direction. For secondary: information collection checklist with found/missing items

"Interview is done" (with transcript or key points)

Stage 2→3 transition. Execute:

1. Organize interview content by theme, tag analysis value (core material / background / uncertain)
2. Propose 2–3 topic directions based on information + theoretical fit, analyzing: importance/timeliness, teaching value/novelty, scope (can 4–6 questions cover it?), information sufficiency
3. Flag information gaps: what additional information is needed for each direction

"Start writing" (or provide existing draft)

Enter Stage 3 iteration loop. From this point, every interaction produces a "triple delivery":

1. Manuscript — current version of case body or teaching note
2. Alignment check:
   • Body section × discussion question coverage matrix
   • Wording consistency scan
   • Theory citation uniformity check
3. Next step suggestion — the single most important thing to advance (e.g., "lock down Q2's theoretical framework" or "Section 3 lacks information to support Q3, suggest supplemental interview on...")

"Topic is locked: [one-sentence description]"

Loop exit. Execute:

1. Re-evaluate body structure against locked topic, suggest adjustments
2. Output complete teaching framework: learning objectives + discussion questions
3. Update coverage matrix, flag sufficient/insufficient areas

"Class has been taught" (with classroom notes)

Stage 4→5. Execute:

1. Integrate classroom feedback into teaching note (discussion highlights, student reactions, instructor commentary)
2. Flag body sections that may need adjustment based on classroom experience

"Ready to submit"

Stage 6. Execute:

1. Rubric benchmarking: Check against rubric item by item (see references/rubric-case-body.md and references/rubric-teaching-note.md), output self-assessment scores and improvement suggestions
2. Full text final check: typos, grammar, logic, timeline, theory citation consistency
3. Format compliance: check against target journal/award requirements

"Reviewer feedback is back" (with review comments)

Execute:

1. Analyze each comment, classify as "substantive revision needed / needs supplemental explanation / polite acknowledgment"
2. Draft revision plan
3. Write Response to Reviewers

Core Principles Inside the Iteration Loop

1. Don't polish details before topic is locked. Get structure and direction right first.
2. Each iteration focuses on one problem. Don't try to fix everything at once.
3. Proactively flag bottlenecks. If information is insufficient for a good topic, say so directly.
4. Theory serves analysis. Literature review finds "the knife that cuts the phenomenon," not citation padding.

Case Difficulty Classification

When planning or reviewing a case, assess its difficulty on three independent dimensions (adapted from Leenders/Erskine/Mauffette-Leenders' Case Difficulty Cube):

Use this to: (1) match case difficulty to target audience (MBA vs. EMBA vs. EE), (2) ensure Exhibits and body structure align with intended difficulty, (3) communicate difficulty to instructors in the teaching note.

This is a planning and communication tool — it does not override the rubric scoring dimensions.

Key Quality Standards (Quick Reference)

Case body (100 points): importance/decision focus (25), complexity/controversy (25), information rigor (25), writing quality (25). See references/rubric-case-body.md.

Teaching note (100 points): learning objectives/question design (25), teaching plan (15), question analysis (35), writing/information support (25). See references/rubric-teaching-note.md.

Writing style: Narrative (HBS + WSJ style). No sensationalism, no consulting jargon, no subjective judgment. See references/writing-style.md.

Teaching note design: What→Why→How question logic, Bloom's taxonomy progression, "升维" (elevation) opportunities. See references/teaching-note-structure.md. For 14-step practical workflow, self-check checklist, scoring reference, and award encouragement directions, see references/practical-workflow.md.

Language Convention

• Chinese cases: body in Chinese, theory terms in English, ~10,000 words (±500)
• English cases: all English, ~8,000 words (±400)
• Follow the user's language for discussion
• APA endnotes for all sources, with hyperlinks in body text

Common Issues

Topic feels generic:

• Check novelty dimension: has this angle been written hundreds of times? Search case libraries again.
• Look for the company's unique differentiation from peers (Line C output).

Writing reads like a news article or consulting report:

• Switch to narrative style: specific time/place/person, dialogue, tension through data contrasts.
• Run the "forbidden expressions" check in references/writing-style.md.

Teaching note questions feel disconnected:

• Check: does each Q build on the previous Q's conclusion?
• Apply What→Why→How logic chain. See references/teaching-note-structure.md.

Information gaps blocking progress:

• Don't force it. Flag the gap explicitly, suggest targeted follow-up interviews or searches.
• Separate "nice to have" from "fatal gaps" (without which no good case can be written).