Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Choose the right e-commerce platform mix for a Taiwan business — DTC platforms (Shopline, 91APP, Shopify), marketplaces (Shopee, momo, PChome), or hybrid. Use when comparing platform fees, traffic potential, brand control trade-offs, or designing a go-to-market channel strategy for Taiwan. Do NOT use for specific platform integration details (see platform-specific skills). STATUS: SKELETON — body pending.
.claude/skills/asgard-ai-platform-tw-ecom-channel-strategy/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-08 | ✓→✓ | ＝ Same ✓ | 3% | 0% |
| case-09 | ✓→✓ | ＝ Same ✓ | 46% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 41% | 0% |
> STATUS: SKELETON — body pending. Prefer tw-ecom-dtc-shopline for platform-specific depth in the meantime.
TODO: decision dimensions (fees, traffic, brand control, ops complexity), DTC vs marketplace typology, platform-fee tiers.
TODO: flow chart keyed on annual revenue / brand strength / ops capacity.
TODO: how to shortlist, pilot, measure, decide.
TODO: 5-6 specific platform-selection traps for Taiwan businesses.
TODO: one non-obvious constraint (candidate: "Platform fees look comparable on paper but traffic acquisition cost is where margins die — always model unit economics with realistic CAC per platform").
TODO: Markdown template for platform selection recommendation.
ecom-rfm-analysis for customer cohort comparison across channels_Last verified: 2026-04_
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | pass→pass | 18,682 | 18,055 | -3% | 1 | 1 | 0% | 3,025 | 3,112 | +3% | 0 | 0 | — |
case-09 | pass→pass | 19,940 | 24,406 | +22% | 1 | 1 | 0% | 2,892 | 4,221 | +46% | 0 | 0 | — |
case-01 | pass→pass | 21,591 | 30,590 | +42% | 1 | 1 | 0% | 3,169 | 4,471 | +41% | 0 | 0 | — |
case-02 | fail→fail | 27,076 | 31,174 | +15% | 1 | 1 | 0% | 3,938 | 6,291 | +60% | 0 | 0 | — |
case-03 | fail→pass | 21,821 | 25,793 | +18% | 1 | 1 | 0% | 3,357 | 4,432 | +32% | 0 | 0 | — |
case-04 | pass→pass | 26,271 | 28,322 | +8% | 1 | 1 | 0% | 3,853 | 5,186 | +35% | 0 | 0 | — |
case-05 | fail→fail | 19,966 | 25,245 | +26% | 1 | 1 | 0% | 3,066 | 4,261 | +39% | 0 | 0 | — |
case-06 | pass→pass | 19,899 | 21,480 | +8% | 1 | 1 | 0% | 2,618 | 3,820 | +46% | 0 | 0 | — |
case-07 | pass→pass | 19,093 | 32,145 | +68% | 1 | 1 | 0% | 3,133 | 4,985 | +59% | 0 | 0 | — |
case-10 | pass→pass | 19,686 | 18,368 | -7% | 1 | 1 | 0% | 2,959 | 3,323 | +12% | 0 | 0 | — |
case-11 | pass→pass | 16,371 | 22,003 | +34% | 1 | 1 | 0% | 2,525 | 3,372 | +34% | 0 | 0 | — |
case-12 | pass→pass | 18,452 | 25,170 | +36% | 1 | 1 | 0% | 2,821 | 4,226 | +50% | 0 | 0 | — |
case-13 | pass→pass | 22,578 | 23,234 | +3% | 1 | 1 | 0% | 3,146 | 3,786 | +20% | 0 | 0 | — |
case-14 | pass→pass | 17,645 | 24,872 | +41% | 1 | 1 | 0% | 2,618 | 3,815 | +46% | 0 | 0 | — |
case-15 | pass→pass | 23,602 | 43,867 | +86% | 1 | 1 | 0% | 3,954 | 7,284 | +84% | 0 | 0 | — |
case-16 | fail→pass | 24,282 | 27,697 | +14% | 1 | 1 | 0% | 3,230 | 4,286 | +33% | 0 | 0 | — |
case-17 | pass→pass | 17,818 | 22,870 | +28% | 1 | 1 | 0% | 2,738 | 3,588 | +31% | 0 | 0 | — |
case-18 | pass→pass | 23,389 | 26,231 | +12% | 1 | 1 | 0% | 3,074 | 3,723 | +21% | 0 | 0 | — |
case-19 | pass→pass | 17,674 | 31,462 | +78% | 1 | 1 | 0% | 3,310 | 5,446 | +65% | 0 | 0 | — |
case-20 | pass→pass | 19,126 | 36,084 | +89% | 1 | 1 | 0% | 3,218 | 5,519 | +72% | 0 | 0 | — |
case-21 | pass→pass | 17,977 | 23,875 | +33% | 1 | 1 | 0% | 3,426 | 4,978 | +45% | 0 | 0 | — |
case-22 | pass→pass | 18,092 | 17,554 | -3% | 1 | 1 | 0% | 2,648 | 2,664 | +1% | 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. 22 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.