Below is a complete, marketplace-ready ChatGPT Agent Skill based on your “Agent Reasoning System” adapted for marketing decision-making, campaign optimization, and strategy iteration. --- Skill Details What are you selling?* Agent Skill Which agent runtime does your skill target? ChatGPT Skill --- Name* Advanced Marketing Decision Engine --- Description* A structured reasoning agent skill that turns marketing decision-making into an iterative optimization system. It continuously refines campaign strategies, audience targeting, messaging, and budget allocation by treating each decision cycle as a state update process. The skill models marketing work as a dynamic system: State (Sₜ): current marketing plan, hypotheses, audience segments, creative variants, KPIs Gradients (Gₜ): performance signals (CTR, CAC, ROAS sensitivity, conversion deltas, engagement shifts) External inputs (Xₜ): market data, A/B test results, constraints, budget limits, competitor signals Update operator (U): logic that adjusts strategy based on signals and constraints This enables the agent to: Continuously optimize campaigns instead of producing static plans Detect weak performance drivers via sensitivity analysis Reallocate budget and messaging dynamically Iterate toward higher ROI marketing configurations Ideal for performance marketers, growth teams, and marketing automation agents. --- Skill File Skill name* advanced-marketing-decision-engine --- When to use* Use when the user asks to optimize, analyze, or iteratively improve marketing campaigns, audience targeting, messaging, or budget allocation using data-driven or performance feedback loops. --- SKILL.md body* --- name: advanced-marketing-decision-engine description: Iterative marketing optimization system using state-update reasoning over campaign data, performance signals, and constraints. version: 1.0.0 --- # Advanced Marketing Decision Engine ## Overview This skill transforms marketing strategy into an iterative optimization system. Every marketing decision is modeled as a state update: Sₜ₊₁ = U(Sₜ, Gₜ, Xₜ) Where: - Sₜ = current marketing system state - Gₜ = gradients (performance sensitivities and diagnostics) - Xₜ = external inputs (data, constraints, experiments) - U = update operator (strategy refinement function) --- ## Core Reasoning Loop At each step: ### 1. Build Current State (Sₜ) Represent the marketing system as structured components: - Audience segments - Campaign structures - Creative variants - Budget allocation - KPI targets ### 2. Compute Gradients (Gₜ) Derive sensitivity signals such as: - CAC elasticity per channel - Conversion rate changes per message variant - ROAS marginal gains per budget shift - Funnel drop-off points - Engagement differentials across segments ### 3. Incorporate External Inputs (Xₜ) Include: - A/B test results - Market changes or seasonality - Budget constraints - Competitive activity - Platform algorithm changes - Customer feedback signals ### 4. Apply Update Operator (U) Update strategy using: - Reallocation of budget toward high-gradient channels - Pruning underperforming segments - Amplifying high-performing creative variants - Adjusting targeting parameters - Updating hypotheses for next iteration --- ## Output Format Always return: ### 1. Current State Summary (Sₜ) Concise representation of current marketing configuration ### 2. Diagnostic Gradients (Gₜ) What is working / not working and why ### 3. External Signals (Xₜ) Key constraints and new data affecting decisions ### 4. Updated Strategy (Sₜ₊₁) Concrete revised marketing plan ### 5. Rationale Explain how gradients and inputs influenced updates --- ## Behavior Rules - Always treat marketing as an iterative optimization system - Prefer measurable signals over intuition alone - Identify causal drivers, not just correlations - Prioritize ROI-maximizing adjustments - Maintain explicit trace of state changes over time - Avoid static “one-shot” marketing plans --- ## Example Workflow User: Improve my paid ads performance for SaaS product Agent: 1. Builds current campaign state (channels, ads, targeting) 2. Computes gradients (CTR drop on LinkedIn, high CAC on Meta) 3. Incorporates external inputs (budget cut, new competitor entry) 4. Updates strategy (shift budget to Google Search, refine messaging) 5. Outputs revised campaign plan --- ## Tools (optional) Read, Edit, Bash --- ## End of Skill --- Examples Example 1* User message (activates the skill) Improve performance of my SaaS Google Ads campaigns. Agent response (what your skill produces) I modeled your campaign as an iterative system and evaluated performance gradients. Sₜ (Current State): Google Search Ads targeting 5 high-intent keywords Mixed ad copy (feature-focused messaging) Budget evenly distributed across all ad groups Primary KPI: CAC Gₜ (Diagnostics): Keyword A: high CTR but low conversion → intent mismatch Keyword B: high conversion, limited volume → under-scaled Ad copy emphasizing features underperforming vs benefit-driven variants Mobile traffic converting 22% lower than desktop Xₜ (External Inputs): Budget constraint reduced by 15% Competitor increased bids on core keyword cluster New landing page variant introduced (not yet tested) Sₜ₊₁ (Updated Strategy): Reallocate 40% of budget to Keyword B cluster Pause lowest-converting keyword sets Shift ad copy toward outcome-based messaging (“reduce churn by 30%”) Split test new landing page against current version Adjust bids downward on mobile traffic segments Rationale: Updates prioritize high-conversion but underfunded segments while reducing spend in low-efficiency areas identified via gradient signals. --- Example 2* User message (activates the skill) Optimize my Instagram ad campaigns for an e-commerce brand. Agent response (what your skill produces) I treated your campaign as a dynamic optimization system and computed performance gradients. Sₜ (Current State): 3 active Instagram campaigns (awareness, traffic, conversion) Broad audience targeting (18–45, lifestyle interest clusters) Creative mix: product shots + influencer content KPI: ROAS Gₜ (Diagnostics): Influencer creatives: highest engagement but low conversion Product shots: stable conversion but declining CTR Audience segment “fitness enthusiasts”: highest ROAS gradient Broad targeting diluting efficiency Xₜ (External Inputs): Holiday season demand spike Rising CPM costs across Meta platforms New influencer content batch available Sₜ₊₁ (Updated Strategy): Increase budget allocation to “fitness enthusiasts” segment Shift influencer content to upper-funnel awareness only Refresh product shot creatives with seasonal positioning Narrow audience targeting to top-performing interest clusters Rebalance spend toward conversion-optimized campaigns Rationale: Strategy shifts concentrate spend where ROAS gradients are strongest while correcting creative mismatch between engagement and conversion layers. --- ChatGPT model version GPT-5.3-mini --- Setup instructions* Install at: ~/.cursor/skills/advanced-marketing-decision-engine/SKILL.md Then in Cursor / ChatGPT Agents: 1. Restart or reload skills registry 2. Ensure skill is enabled in agent settings 3. Trigger with marketing optimization prompts (ads, campaigns, growth strategy) --- If you want, I can also: convert this into a **multi-agent system (planner + optimizer + evaluator)** or extend it into a **real-time A/B testing automation skill with tooling hooks**