Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Plan and review ChatGPT Ads experiments without assuming ad inventory, account access, or platform capabilities. Use for ChatGPT advertising, OpenAI ads, conversational ad concepts, ChatGPT campaign plans, or performance reviews from a verified export.
.claude/skills/nowork-studio-paid-ads-chatgpt/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-22 | ✓→✗ | ▼ Worse | -12% | 0% |
| case-20 | ✓→✓ | ＝ Same ✓ | 16% | 0% |
| case-21 | ✓→✓ | ＝ Same ✓ | 7% | 0% |
| case-23 | ✓→✓ | ＝ Same ✓ | 61% | 0% |
Read ../../paid-ads/paid-ads-chatgpt/SKILL.md completely, then follow it as the active workflow. Normalize that path from the directory containing this wrapper: the canonical file is <plugin-root>/paid-ads/paid-ads-chatgpt/SKILL.md, not <plugin-root>/skills/paid-ads-chatgpt/paid-ads-chatgpt/SKILL.md. Resolve every relative reference from the canonical file against <plugin-root>/paid-ads/paid-ads-chatgpt/. If the canonical file cannot be read, stop and report the packaging error; never substitute a similarly named skill from another plugin.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 25,602 | 46,073 | +80% | 1 | 1 | 0% | 4,098 | 4,182 | +2% | 0 | 0 | — |
case-02 | fail→fail | 18,361 | 24,508 | +33% | 1 | 1 | 0% | 2,757 | 3,544 | +29% | 0 | 0 | — |
case-03 | fail→pass | 14,217 | 7,622 | -46% | 1 | 1 | 0% | 2,788 | 1,276 | -54% | 0 | 0 | — |
case-04 | fail→fail | 13,802 | 13,960 | +1% | 1 | 1 | 0% | 2,194 | 2,218 | +1% | 0 | 0 | — |
case-05 | fail→fail | 24,969 | 22,982 | -8% | 1 | 1 | 0% | 2,839 | 3,273 | +15% | 0 | 0 | — |
case-06 | fail→fail | 17,192 | 20,413 | +19% | 1 | 1 | 0% | 2,293 | 3,133 | +37% | 0 | 0 | — |
case-07 | fail→fail | 13,004 | 16,613 | +28% | 1 | 1 | 0% | 2,109 | 2,902 | +38% | 0 | 0 | — |
case-08 | fail→fail | 14,054 | 15,203 | +8% | 1 | 1 | 0% | 2,123 | 2,233 | +5% | 0 | 0 | — |
case-09 | fail→fail | 12,250 | 25,593 | +109% | 1 | 1 | 0% | 1,911 | 2,913 | +52% | 0 | 0 | — |
case-10 | fail→fail | 16,097 | 21,723 | +35% | 1 | 1 | 0% | 2,709 | 4,002 | +48% | 0 | 0 | — |
case-11 | fail→fail | 17,272 | 21,964 | +27% | 1 | 1 | 0% | 2,477 | 3,204 | +29% | 0 | 0 | — |
case-12 | fail→fail | 19,459 | 15,717 | -19% | 1 | 1 | 0% | 1,714 | 2,609 | +52% | 0 | 0 | — |
case-13 | fail→fail | 14,264 | 29,008 | +103% | 1 | 1 | 0% | 2,224 | 2,802 | +26% | 0 | 0 | — |
case-14 | fail→fail | 10,437 | 25,808 | +147% | 1 | 1 | 0% | 1,607 | 1,881 | +17% | 0 | 0 | — |
case-15 | fail→fail | 19,476 | 12,866 | -34% | 1 | 1 | 0% | 1,545 | 2,014 | +30% | 0 | 0 | — |
case-16 | fail→fail | 14,188 | 11,183 | -21% | 1 | 1 | 0% | 1,954 | 1,939 | -1% | 0 | 0 | — |
case-17 | fail→fail | 12,338 | 26,937 | +118% | 1 | 1 | 0% | 1,804 | 2,565 | +42% | 0 | 0 | — |
case-18 | fail→fail | 15,335 | 16,352 | +7% | 1 | 1 | 0% | 2,273 | 2,552 | +12% | 0 | 0 | — |
case-19 | fail→fail | 15,823 | 21,719 | +37% | 1 | 1 | 0% | 2,585 | 3,845 | +49% | 0 | 0 | — |
case-20 | pass→pass | 9,516 | 18,509 | +95% | 1 | 1 | 0% | 1,343 | 1,556 | +16% | 0 | 0 | — |
case-21 | pass→pass | 13,922 | 16,796 | +21% | 1 | 1 | 0% | 2,141 | 2,298 | +7% | 0 | 0 | — |
case-22 | pass→fail | 20,497 | 14,178 | -31% | 1 | 1 | 0% | 2,775 | 2,449 | -12% | 0 | 0 | — |
case-23 | pass→pass | 10,476 | 16,050 | +53% | 1 | 1 | 0% | 1,580 | 2,551 | +61% | 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 0 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/29/2026 | +32% |
Other measured skills in the registry, with their headline benchmark lift.