Evaluation
AuraScore 81/100

Dynamic Pricing Agent Safety and Margin Risk Assessment Memo

Evaluate autonomous retail pricing agents against gross margin thresholds, competitor tracking errors, and compliance guardrails.

Run this evaluation when automated algorithmic pricing agents demonstrate volatility, margin erosion, or unintended promotional stacking. It delivers a high-impact risk memo to merchandising and pricing executives.

Template

Role: Director of Algorithmic Commerce Evaluation and Margin Protection with deep expertise in retail dynamic pricing agent systems.

Context

  • Retail enterprise: {{retail_chain_name}}
  • Pricing agent architecture and version: {{pricing_engine_tier}}
  • Target product categories evaluated: {{evaluated_sku_categories}}
  • Gross margin and variance telemetry: {{margin_variance_data}}
  • Guardrail violation and price floor override logs: {{guardrail_violation_logs}}
  • Net revenue and volume elasticity impact: {{revenue_impact_metrics}}

Task

Draft an executive-level evaluation email to the Chief Merchandising Officer detailing the operational safety, margin preservation integrity, and algorithmic stability of the dynamic pricing agent based on recent production telemetry.

Method

  1. Ingest {{margin_variance_data}} to determine if the pricing agent's autonomous repricing actions remained within mandated gross margin corridors.
  2. Audit {{guardrail_violation_logs}} for edge-case failures including competitive price spiral loops, stale inventory scrapers, and cross-promotional price stacking.
  3. Quantify the net contribution margin impact across {{evaluated_sku_categories}} using {{revenue_impact_metrics}}.
  4. Evaluate agent reward function convergence to verify whether volume maximization is cannibalizing high-margin premium SKUs.
  5. Benchmark the agent's real-time latency when reacting to sudden supplier cost spikes versus competitor out-of-stock events.
  6. Classify system risk level (Low, Moderate, Critical) with strict reference to price-gouging compliance and MAP (Minimum Advertised Price) adherence.
  7. Propose three hard-enforcement kill-switches and boundary re-calibrations for immediate deployment in {{pricing_engine_tier}}.

Constraints

  • MUST include explicit financial trade-offs (margin dollar variance vs. unit sales velocity).
  • MUST NOT exceed 600 words across the entire email.
  • MUST classify risk clearly into an executive rating banner at the very top of the email.
  • Every technical critique MUST map directly to commercial merchandising impact.

Output format

Email structure:

  • Subject Line (standardized: [PRICING AGENT AUDIT] Risk & Margin Variance Report - {{retail_chain_name}})
  • Risk Classification Banner (e.g., RISK LEVEL: MODERATE | ACTION REQUIRED)
  • Executive Summary (Strategic overview of pricing agent health)
  • Key Telemetry & Margin Divergence Analysis (Categorized findings from {{evaluated_sku_categories}})
  • Policy & MAP Compliance Violations (Breakdown of {{guardrail_violation_logs}})
  • Required Algorithmic Guardrails (3 mandatory system adjustments)
  • Merchandising Sign-Off & Rollback Conditions

Self-review

  • Verify that both margin erosion and price floor compliance are explicitly evaluated.
  • Ensure all variables ({{retail_chain_name}}, {{margin_variance_data}}, etc.) are accurately reflected.
  • Confirm the email length is strictly between 400 and 600 words.
AuraScore breakdown
81/100Provisional
Instruction clarity15/15 · Strong

Explicit role, a named task, and discrete steps the model can follow.

Context architecture12/12 · Strong

Background, inputs and variables the model needs before it starts.

Constraint engineering12/12 · Strong

Hard boundaries — what the model must and must not do.

Output specification6/14 · Thin

A named, field-level shape for the response.

Reasoning structure10/10 · Strong

Ordered work items that force analysis before an answer.

Model compatibility10/10 · Strong

Length and structure that travel across frontier models.

Token efficiency5/10 · Thin

Signal density — instruction weight without padding.

Reusability7/7 · Strong

Documented variables so the scaffold adapts to new inputs.

Robustness3/5 · Adequate

Quality bar, assumptions and behaviour when inputs are thin.

Observed performance1/5 · Thin

How much real usage the template has behind it.

ai-agents
agents-evaluation
retail-consumer-goods
retail
pricing-agents
algorithmic-commerce