Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Apply Institutional Theory (DiMaggio and Powell, 1983) to analyze how coercive, mimetic, and normative isomorphic pressures shape organizational structures and practices. Use this skill when the user needs to explain why organizations in the same field look alike, evaluate whether a practice was adopted for legitimacy vs efficiency, analyze regulatory or social pressures on strategy, or when they ask 'why do all firms in this industry do the same thing', 'is this best practice or just conformity
.claude/skills/asgard-ai-platform-grad-strat-institutional/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-20 | ✓→✗ | ▼ Worse | 40% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | -46% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | -36% | 0% |
Institutional theory explains why organizations within the same field become structurally similar (isomorphism) — not necessarily because the adopted practices are efficient, but because institutional pressures demand conformity for legitimacy and survival. DiMaggio and Powell (1983) identified three isomorphic mechanisms.
IRON LAW: Organizations may adopt practices for LEGITIMACY, not
EFFICIENCY. Assuming all organizational practices are efficiency-
driven will misdiagnose the real adoption motive and lead to
incorrect strategic recommendations.Key assumptions:
| Mechanism | Source | Driver | Example | |-----------|--------|--------|---------| | Coercive | Regulations, laws, mandates from powerful actors | Compliance with authority | Government mandating ESG reporting | | Mimetic | Uncertainty; copying successful/prominent organizations | Uncertainty reduction | Startups copying FAANG organizational structures | | Normative | Professionalization, education, professional networks | Professional standards | MBA programs teaching identical frameworks |
Decoupling occurs when organizations formally adopt practices (for legitimacy) but decouple them from actual operations. Indicators:
markdown## Institutional Analysis: [Context] ### Organizational Field Definition - Field boundaries: ... - Key actors: ... ### Isomorphic Pressures | Practice/Structure | Mechanism | Source | Motive (Efficiency/Legitimacy) | |-------------------|-----------|--------|-------------------------------| | [practice] | [C/M/N] | [source] | [motive] | ### Decoupling Assessment - Formally adopted but decoupled: ... - Tightly coupled (genuine adoption): ... ### Strategic Implications 1. [respond to coercive pressures: compliance strategy] 2. [respond to mimetic pressures: differentiation vs conformity] 3. [respond to normative pressures: professional development]
Analyzing why all firms in a sector adopted CSR reporting: coercive (regulatory mandate in some jurisdictions), mimetic (industry leaders published reports, others followed under uncertainty), normative (business school curricula emphasize stakeholder management). Assessment finds most firms decouple — reports exist but practices are ceremonial.
Claiming "firms adopt CSR because it's profitable" without considering institutional pressures. Institutional theory specifically warns against assuming efficiency motives for all organizational practices.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 30,869 | 19,552 | -37% | 1 | 1 | 0% | 4,731 | 4,047 | -14% | 0 | 0 | — |
case-02 | pass→pass | 57,099 | 23,777 | -58% | 1 | 1 | 0% | 8,316 | 4,463 | -46% | 0 | 0 | — |
case-03 | pass→pass | 40,469 | 20,359 | -50% | 1 | 1 | 0% | 6,505 | 4,141 | -36% | 0 | 0 | — |
case-04 | pass→pass | 14,043 | 17,004 | +21% | 1 | 1 | 0% | 1,957 | 3,514 | +80% | 0 | 0 | — |
case-05 | pass→pass | 17,692 | 11,740 | -34% | 1 | 1 | 0% | 2,702 | 2,762 | +2% | 0 | 0 | — |
case-11 | fail→pass | 17,102 | 14,616 | -15% | 1 | 1 | 0% | 2,468 | 3,178 | +29% | 0 | 0 | — |
case-06 | pass→pass | 16,177 | 11,672 | -28% | 1 | 1 | 0% | 2,313 | 2,745 | +19% | 0 | 0 | — |
case-07 | pass→pass | 15,592 | 12,854 | -18% | 1 | 1 | 0% | 2,489 | 3,026 | +22% | 0 | 0 | — |
case-08 | pass→pass | 16,136 | 11,518 | -29% | 1 | 1 | 0% | 2,356 | 2,762 | +17% | 0 | 0 | — |
case-09 | pass→pass | 15,237 | 11,517 | -24% | 1 | 1 | 0% | 2,396 | 2,780 | +16% | 0 | 0 | — |
case-10 | pass→pass | 13,001 | 13,632 | +5% | 1 | 1 | 0% | 2,078 | 3,106 | +49% | 0 | 0 | — |
case-12 | fail→fail | 20,208 | 11,833 | -41% | 1 | 1 | 0% | 3,209 | 2,716 | -15% | 0 | 0 | — |
case-13 | pass→pass | 18,254 | 12,226 | -33% | 1 | 1 | 0% | 2,838 | 2,869 | +1% | 0 | 0 | — |
case-14 | pass→pass | 11,642 | 10,023 | -14% | 1 | 1 | 0% | 1,651 | 2,534 | +53% | 0 | 0 | — |
case-15 | pass→pass | 17,384 | 11,342 | -35% | 1 | 1 | 0% | 2,501 | 2,480 | -1% | 0 | 0 | — |
case-16 | pass→pass | 16,597 | 12,360 | -26% | 1 | 1 | 0% | 2,574 | 2,931 | +14% | 0 | 0 | — |
case-17 | pass→pass | 29,850 | 20,965 | -30% | 1 | 1 | 0% | 4,337 | 3,955 | -9% | 0 | 0 | — |
case-18 | pass→pass | 21,243 | 17,070 | -20% | 1 | 1 | 0% | 3,030 | 3,512 | +16% | 0 | 0 | — |
case-19 | pass→pass | 16,209 | 15,421 | -5% | 1 | 1 | 0% | 2,221 | 3,073 | +38% | 0 | 0 | — |
case-20 | pass→fail | 21,165 | 22,660 | +7% | 1 | 1 | 0% | 3,295 | 4,624 | +40% | 0 | 0 | — |
case-21 | pass→pass | 21,613 | 22,843 | +6% | 1 | 1 | 0% | 3,272 | 4,281 | +31% | 0 | 0 | — |
case-22 | pass→pass | 13,282 | 12,211 | -8% | 1 | 1 | 0% | 2,042 | 3,203 | +57% | 0 | 0 | — |
case-23 | pass→pass | 14,675 | 27,271 | +86% | 1 | 1 | 0% | 2,885 | 7,191 | +149% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +4 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.