Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Strategic paid advertising planning with industry-specific templates. Covers platform selection, campaign architecture, budget planning, creative strategy, and phased implementation roadmap. Triggers on: "ad plan", "ad strategy", "campaign planning", "media plan", "PPC strategy", "advertising plan", "media strategy"
.claude/skills/miosa-osa-ads-plan/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 51% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 72% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 146% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 80% | 0% |
> End-to-end advertising strategy with platform selection, campaign architecture, > budget planning, creative strategy, and phased implementation roadmap.
bash/ads plan /ads plan --industry saas /ads plan --budget 10000
Match business type to recommended platform mix:
| Business Type | Primary | Secondary | Testing | |---------------|---------|-----------|---------| | SaaS B2B | Google Search, LinkedIn | Meta, YouTube | TikTok, Microsoft | | E-commerce | Google Shopping, Meta | TikTok, YouTube | Microsoft, LinkedIn | | Local Service | Google Search, Google LSA | Meta | Microsoft, YouTube | | B2B Enterprise | LinkedIn, Google Search | Meta | Microsoft, TikTok | | Info Products | Meta, YouTube | Google Search | TikTok | | Mobile App | Meta, Google UAC | TikTok | Apple Search Ads |
Naming Convention: [Platform]_[Objective]_[Audience]_[Geo]_[Date]
Account
-- Brand Campaign (always-on)
-- Non-Brand Prospecting
-- Top Funnel (Awareness)
-- Mid Funnel (Consideration)
-- Bottom Funnel (Conversion)
-- Retargeting
-- Website Visitors (7-30 days)
-- Engaged Users
-- Cart Abandoners / Form Starters
-- Testing
-- New audiences, formats, messaging| Tier | Allocation | Purpose | |------|-----------|---------| | Proven (70%) | Primary platforms with proven ROI | Revenue engine | | Scaling (20%) | Platforms showing promise | Growth engine | | Testing (10%) | New platforms or strategies | Innovation |
| Priority | Asset Type | Platforms | Quantity | |----------|-----------|-----------|----------| | P1 | Product/service videos (15-30s) | Meta, TikTok, YouTube | 5-10 | | P2 | Static images with copy | Google, Meta, LinkedIn | 10-15 | | P3 | Carousel/collection | Meta, LinkedIn | 3-5 | | P4 | UGC/testimonial video | TikTok, Meta | 3-5 | | P5 | Long-form video (1-3 min) | YouTube | 2-3 |
| Platform | Client-Side | Server-Side | Priority | |----------|------------|-------------|----------| | Google | gtag.js | Enhanced Conversions, GTM SS | P1 | | Meta | Pixel | CAPI | P1 | | LinkedIn | Insight Tag | CAPI (2025) | P2 | | TikTok | Pixel | Events API + ttclid | P2 | | Microsoft | UET Tag | Enhanced Conversions | P2 |
ADS-STRATEGY.md — Complete strategic advertising planCAMPAIGN-ARCHITECTURE.md — Campaign structure with naming conventionsBUDGET-PLAN.md — Budget allocation with monthly pacingCREATIVE-BRIEF.md — Creative production planTRACKING-SETUP.md — Tracking implementation checklistIMPLEMENTATION-ROADMAP.md — Phased rollout timeline| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 53,859 | 38,241 | -29% | 1 | 1 | 0% | 8,263 | 6,588 | -20% | 0 | 0 | — |
case-02 | fail→pass | 47,856 | 54,430 | +14% | 1 | 1 | 0% | 8,261 | 9,341 | +13% | 0 | 0 | — |
case-03 | fail→fail | 34,584 | 33,308 | -4% | 1 | 1 | 0% | 5,554 | 6,106 | +10% | 0 | 0 | — |
case-04 | pass→pass | 14,693 | 15,963 | +9% | 1 | 1 | 0% | 2,168 | 3,390 | +56% | 0 | 0 | — |
case-05 | pass→pass | 36,958 | 45,536 | +23% | 1 | 1 | 0% | 8,216 | 9,296 | +13% | 0 | 0 | — |
case-06 | pass→pass | 18,456 | 21,719 | +18% | 1 | 1 | 0% | 2,569 | 4,118 | +60% | 0 | 0 | — |
case-07 | pass→pass | 19,856 | 33,314 | +68% | 1 | 1 | 0% | 3,198 | 5,914 | +85% | 0 | 0 | — |
case-17 | fail→pass | 32,742 | 54,670 | +67% | 1 | 1 | 0% | 6,187 | 9,314 | +51% | 0 | 0 | — |
case-08 | pass→pass | 19,936 | 16,934 | -15% | 1 | 1 | 0% | 2,943 | 3,777 | +28% | 0 | 0 | — |
case-09 | fail→fail | 18,581 | 16,732 | -10% | 1 | 1 | 0% | 2,723 | 3,474 | +28% | 0 | 0 | — |
case-10 | fail→fail | 13,717 | 20,721 | +51% | 1 | 1 | 0% | 2,282 | 4,819 | +111% | 0 | 0 | — |
case-11 | pass→pass | 20,501 | 30,826 | +50% | 1 | 1 | 0% | 3,313 | 6,282 | +90% | 0 | 0 | — |
case-12 | pass→pass | 13,614 | 18,236 | +34% | 1 | 1 | 0% | 2,362 | 4,343 | +84% | 0 | 0 | — |
case-13 | fail→pass | 16,456 | 22,674 | +38% | 1 | 1 | 0% | 2,600 | 4,470 | +72% | 0 | 0 | — |
case-14 | fail→fail | 20,650 | 14,200 | -31% | 1 | 1 | 0% | 3,356 | 3,484 | +4% | 0 | 0 | — |
case-15 | pass→pass | 35,858 | 20,780 | -42% | 1 | 1 | 0% | 5,506 | 4,884 | -11% | 0 | 0 | — |
case-16 | fail→pass | 21,242 | 43,320 | +104% | 1 | 1 | 0% | 3,409 | 8,392 | +146% | 0 | 0 | — |
case-18 | fail→pass | 14,634 | 20,833 | +42% | 1 | 1 | 0% | 2,564 | 4,604 | +80% | 0 | 0 | — |
case-19 | pass→pass | 15,548 | 15,369 | -1% | 1 | 1 | 0% | 2,258 | 3,653 | +62% | 0 | 0 | — |
case-20 | fail→pass | 15,462 | 10,447 | -32% | 1 | 1 | 0% | 2,429 | 2,931 | +21% | 0 | 0 | — |
case-21 | pass→pass | 16,612 | 18,543 | +12% | 1 | 1 | 0% | 2,607 | 4,021 | +54% | 0 | 0 | — |
case-22 | pass→pass | 18,212 | 19,560 | +7% | 1 | 1 | 0% | 2,362 | 3,803 | +61% | 0 | 0 | — |
case-23 | pass→pass | 14,279 | 11,648 | -18% | 1 | 1 | 0% | 2,081 | 2,911 | +40% | 0 | 0 | — |
case-24 | fail→pass | 14,623 | 16,568 | +13% | 1 | 1 | 0% | 2,130 | 3,364 | +58% | 0 | 0 | — |
case-25 | pass→pass | 17,469 | 20,167 | +15% | 1 | 1 | 0% | 2,737 | 3,902 | +43% | 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. 25 cases were attempted. The headline lift of +28 percentage points is the difference between those two pass rates over the 25 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.