Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →YouTube Ads specific analysis covering campaign types, creative quality, audience targeting, and measurement. Evaluates video ad performance across skippable, non-skippable, bumper, Shorts, and Demand Gen formats. Triggers on: "YouTube Ads", "video ads", "pre-roll", "bumper ads", "YouTube campaign", "Shorts ads", "YouTube audit"
.claude/skills/miosa-osa-ads-youtube/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 18% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 46% | 0% |
> Campaign-by-campaign analysis of YouTube advertising across all video formats.
bash/ads youtube
| Metric | Benchmark | Notes | |--------|-----------|-------| | View Rate (skippable) | >=15% | Higher = better hook | | CPV (skippable) | $0.01-0.10 | Varies by targeting | | VTR (bumper) | 90%+ | Non-skippable | | CPM (non-skip) | $6-15 | Varies by market | | CTR (Demand Gen) | >=0.5% | Image+video combined |
| Category | Weight | |----------|--------| | Creative Quality | 30% | | Campaign Setup | 25% | | Audience Targeting | 25% | | Measurement | 20% |
YOUTUBE-ADS-REPORT.md — Campaign-by-campaign analysis| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | pass→pass | 19,023 | 22,904 | +20% | 1 | 1 | 0% | 2,522 | 4,659 | +85% | 0 | 0 | — |
case-01 | fail→pass | 101,030 | 35,487 | -65% | 1 | 1 | 0% | 8,251 | 7,120 | -14% | 0 | 0 | — |
case-02 | fail→pass | 36,403 | 35,766 | -2% | 1 | 1 | 0% | 5,984 | 6,984 | +17% | 0 | 0 | — |
case-03 | fail→fail | 44,932 | 37,086 | -17% | 1 | 1 | 0% | 7,159 | 7,299 | +2% | 0 | 0 | — |
case-04 | pass→pass | 18,904 | 21,267 | +13% | 1 | 1 | 0% | 2,980 | 3,689 | +24% | 0 | 0 | — |
case-05 | fail→fail | 19,183 | 16,615 | -13% | 1 | 1 | 0% | 2,576 | 3,623 | +41% | 0 | 0 | — |
case-06 | fail→pass | 13,857 | 12,062 | -13% | 1 | 1 | 0% | 2,283 | 2,691 | +18% | 0 | 0 | — |
case-07 | fail→pass | 21,216 | 22,969 | +8% | 1 | 1 | 0% | 3,095 | 3,991 | +29% | 0 | 0 | — |
case-08 | pass→pass | 13,377 | 20,625 | +54% | 1 | 1 | 0% | 2,292 | 3,251 | +42% | 0 | 0 | — |
case-09 | fail→fail | 12,520 | 18,319 | +46% | 1 | 1 | 0% | 2,064 | 3,392 | +64% | 0 | 0 | — |
case-11 | fail→pass | 15,301 | 13,954 | -9% | 1 | 1 | 0% | 2,105 | 3,069 | +46% | 0 | 0 | — |
case-12 | pass→pass | 38,104 | 23,671 | -38% | 1 | 1 | 0% | 2,900 | 3,980 | +37% | 0 | 0 | — |
case-13 | pass→pass | 14,959 | 16,137 | +8% | 1 | 1 | 0% | 2,113 | 2,956 | +40% | 0 | 0 | — |
case-14 | fail→pass | 12,802 | 15,966 | +25% | 1 | 1 | 0% | 2,133 | 3,666 | +72% | 0 | 0 | — |
case-15 | fail→pass | 19,331 | 11,251 | -42% | 1 | 1 | 0% | 2,617 | 2,821 | +8% | 0 | 0 | — |
case-16 | pass→pass | 16,145 | 25,988 | +61% | 1 | 1 | 0% | 2,507 | 2,936 | +17% | 0 | 0 | — |
case-17 | fail→pass | 20,730 | 12,397 | -40% | 1 | 1 | 0% | 2,974 | 3,104 | +4% | 0 | 0 | — |
case-18 | pass→pass | 16,328 | 17,453 | +7% | 1 | 1 | 0% | 2,362 | 3,547 | +50% | 0 | 0 | — |
case-19 | fail→pass | 18,303 | 17,929 | -2% | 1 | 1 | 0% | 2,680 | 3,643 | +36% | 0 | 0 | — |
case-20 | pass→pass | 11,528 | 14,088 | +22% | 1 | 1 | 0% | 1,637 | 2,818 | +72% | 0 | 0 | — |
case-21 | pass→pass | 18,476 | 24,994 | +35% | 1 | 1 | 0% | 1,993 | 5,033 | +153% | 0 | 0 | — |
case-22 | pass→pass | 21,065 | 27,937 | +33% | 1 | 1 | 0% | 2,755 | 5,636 | +105% | 0 | 0 | — |
case-23 | pass→pass | 19,742 | 27,568 | +40% | 1 | 1 | 0% | 2,716 | 5,808 | +114% | 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 +39 percentage points is the difference between those two pass rates over the 23 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.