Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Apply the Uppsala Internationalization Model to analyze gradual foreign market entry based on psychic distance and experiential learning. Use this skill when the user needs to plan a staged internationalization sequence, understand why firms enter culturally similar markets first, or evaluate whether a firm's international expansion follows the establishment chain from export to subsidiary.
.claude/skills/asgard-ai-platform-grad-uppsala/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 56% | 0% |
| case-06 | ✓→✗ | ▼ Worse | 22% | 0% |
| case-07 | ✓→✗ | ▼ Worse | 47% | 0% |
The Uppsala Model explains internationalization as a gradual, path-dependent process driven by experiential learning. Firms enter foreign markets incrementally, starting with psychically close countries (similar language, culture, institutions) and progressing through an establishment chain: no regular export, export via agent, sales subsidiary, production subsidiary. Each stage builds market knowledge that enables the next commitment.
Trigger conditions:
When NOT to use:
IRON LAW: Internationalization Is Path-Dependent
Firms enter PSYCHICALLY CLOSE markets first — those with similar
language, culture, political systems, and business practices.
Each market entry builds experiential knowledge that reduces the
perceived risk of entering the NEXT, more distant market.
Skipping stages (e.g., jumping from no exports to production
subsidiary in a distant market) violates the model and dramatically
increases failure risk — UNLESS the firm has compensating mechanisms
(acquisitions, network relationships, prior international experience).For each market the firm operates in, document:
Assess psychic distance from the home market across: language, culture (Hofstede/GLOBE), political system, economic development, and business practices. Create a ranking: Near, Moderate, Far.
For each target market, plan the progression:
| Stage | Activity | Knowledge Required | Commitment Level | |-------|----------|-------------------|-----------------| | 1 | No regular export | Minimal market awareness | None | | 2 | Export via independent agent | Basic market demand knowledge | Low | | 3 | Sales subsidiary | Customer relationships, distribution knowledge | Medium | | 4 | Production subsidiary | Deep operational knowledge, local supply chains | High |
Specify trigger conditions for advancing to the next stage (e.g., revenue threshold, relationship maturity).
The Uppsala model is a dynamic feedback loop:
Design explicit learning mechanisms: expatriate rotations, local hires, partnership structures.
markdown# Uppsala Internationalization Plan: {Firm} ## Current Footprint | Market | Mode | Years | Knowledge | Commitment | |--------|------|-------|-----------|------------| | {Market A} | {mode} | {N} | {L/M/H} | {L/M/H} | ## Psychic Distance Ranking from {Home Country} - Near: {markets} | Moderate: {markets} | Far: {markets} ## Recommended Entry Sequence 1. {Market}: {mode} -> {next mode} (trigger: {condition}) ## Learning Mechanisms - {Mechanism}: {how it builds market knowledge}
references/uppsala-2009-revision.mdreferences/psychic-distance-scales.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 46,087 | 25,752 | -44% | 1 | 1 | 0% | 7,321 | 4,466 | -39% | 0 | 0 | — |
case-02 | fail→pass | 38,446 | 21,325 | -45% | 1 | 1 | 0% | 5,446 | 4,340 | -20% | 0 | 0 | — |
case-03 | fail→fail | 38,518 | 25,207 | -35% | 1 | 1 | 0% | 6,293 | 4,988 | -21% | 0 | 0 | — |
case-04 | pass→pass | 16,487 | 17,324 | +5% | 1 | 1 | 0% | 2,629 | 3,826 | +46% | 0 | 0 | — |
case-05 | pass→pass | 11,317 | 14,333 | +27% | 1 | 1 | 0% | 1,857 | 3,047 | +64% | 0 | 0 | — |
case-06 | pass→fail | 18,690 | 14,426 | -23% | 1 | 1 | 0% | 2,681 | 3,276 | +22% | 0 | 0 | — |
case-07 | pass→fail | 16,030 | 16,448 | +3% | 1 | 1 | 0% | 2,417 | 3,559 | +47% | 0 | 0 | — |
case-08 | fail→pass | 17,242 | 16,077 | -7% | 1 | 1 | 0% | 2,741 | 3,491 | +27% | 0 | 0 | — |
case-09 | pass→pass | 20,879 | 18,329 | -12% | 1 | 1 | 0% | 2,837 | 3,659 | +29% | 0 | 0 | — |
case-10 | pass→pass | 22,084 | 20,349 | -8% | 1 | 1 | 0% | 3,013 | 3,939 | +31% | 0 | 0 | — |
case-11 | pass→pass | 17,603 | 18,449 | +5% | 1 | 1 | 0% | 2,381 | 3,713 | +56% | 0 | 0 | — |
case-12 | pass→pass | 21,045 | 21,531 | +2% | 1 | 1 | 0% | 2,864 | 4,131 | +44% | 0 | 0 | — |
case-13 | pass→pass | 16,024 | 16,636 | +4% | 1 | 1 | 0% | 2,530 | 3,553 | +40% | 0 | 0 | — |
case-14 | pass→pass | 18,400 | 15,710 | -15% | 1 | 1 | 0% | 2,687 | 3,395 | +26% | 0 | 0 | — |
case-15 | pass→pass | 15,671 | 13,112 | -16% | 1 | 1 | 0% | 2,215 | 3,118 | +41% | 0 | 0 | — |
case-16 | pass→pass | 15,332 | 17,855 | +16% | 1 | 1 | 0% | 2,140 | 3,468 | +62% | 0 | 0 | — |
case-17 | pass→pass | 18,686 | 17,349 | -7% | 1 | 1 | 0% | 2,706 | 3,824 | +41% | 0 | 0 | — |
case-18 | pass→pass | 15,547 | 16,280 | +5% | 1 | 1 | 0% | 2,338 | 3,356 | +44% | 0 | 0 | — |
case-19 | pass→pass | 17,980 | 17,942 | -0% | 1 | 1 | 0% | 2,787 | 3,827 | +37% | 0 | 0 | — |
case-20 | pass→pass | 17,433 | 14,316 | -18% | 1 | 1 | 0% | 2,445 | 2,995 | +22% | 0 | 0 | — |
case-21 | fail→pass | 9,132 | 6,691 | -27% | 1 | 1 | 0% | 1,340 | 2,091 | +56% | 0 | 0 | — |
case-22 | pass→pass | 25,363 | 23,824 | -6% | 1 | 1 | 0% | 3,427 | 4,185 | +22% | 0 | 0 | — |
case-23 | pass→pass | 18,263 | 18,397 | +1% | 1 | 1 | 0% | 2,876 | 3,983 | +38% | 0 | 0 | — |
case-24 | pass→pass | 7,667 | 7,970 | +4% | 1 | 1 | 0% | 1,250 | 2,348 | +88% | 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. 24 cases were attempted. The headline lift of +4 percentage points is the difference between those two pass rates over the 24 comparable cases. 2 cases got worse with the skill loaded, and they are 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.