icaire-content-optimization

Convert validated ICAIRE social-performance findings into next-week content actions and testable hypotheses when planning an optimization cycle.

SKILL.md

GitHub

ICAIRE Content Optimization

Turn evidence-labeled ICAIRE social findings into a focused, testable next-week content plan.

Contract Checklist

  • Read the canonical ICAIRE Brain page protocol/icaire-content-optimization-interpretation before interpreting findings.
  • Accept only traceable findings produced under the social-performance analysis protocol.
  • Preserve each finding's evidence label, platform scope, sample size, and limitations.
  • Produce a report with exactly two main sections: Analysis and Optimization.
  • Make next-week actions specific, feasible, review-gated, and tied to named evidence or hypotheses.
  • Separate exploitation of stronger observed patterns from experiments designed to test weak signals.
  • Specify what will be measured, the comparison, the decision rule, and the future review date.
  • Do not draft final posts, publish, message, or modify account state.

Workflow

  1. Load the protocol and analysis handoff:
    • Read protocol/icaire-content-optimization-interpretation from the live ICAIRE Brain.
    • Verify the analysis artifact, source run IDs, evidence labels, platform scope, and limitations.
    • Stop if findings are not traceable or the protocol is unavailable.
    • Anti-patterns: optimizing from an aggregate PDF alone, using remembered protocol text, accepting unlabeled claims as findings
  2. Restate the defensible analysis:
    • Build the Analysis section from validated findings only.
    • Lead with cross-cutting patterns when supported, then platform-specific signals, counterexamples, and important uncertainties.
    • Keep data-quality issues visible and distinguish observed patterns from weak signals.
    • Anti-patterns: hiding caveats, rewriting weak signals as conclusions, repeating every metric without interpretation
  3. Translate findings into next-week choices:
    • Build the Optimization section as a prioritized set of content actions for the next seven-day cycle.
    • Tie each action to a finding identifier or label it as a new hypothesis.
    • Specify target platform, topic or message angle, language, timing window, format, visual treatment, owner, and review gate when known.
    • Keep the plan small enough to execute and learn from.
    • Anti-patterns: generic advice such as post more, unsupported prescriptions, changing several uncontrolled dimensions in every post
  4. Design testable hypotheses:
    • Express each test as If [controlled content choice], then [metric] will [direction] versus [comparison], because [evidence or rationale].
    • Define the primary metric, comparison or baseline, minimum observation window, confounders to hold stable, and decision rule.
    • Pre-register hypotheses before publishing and do not alter success criteria after seeing results.
    • Anti-patterns: unmeasurable hypotheses, outcome switching, tests without a comparison, causal certainty from observational history
  5. Check portfolio balance and safeguards:
    • Balance exploitation of stronger patterns with a limited number of learning experiments.
    • Respect ICAIRE public-communications approval, confidentiality, accessibility, Arabic and English quality, and brand requirements.
    • Flag any action needing team confirmation or missing source evidence.
    • Anti-patterns: exposing internal-only work, optimizing only for vanity metrics, bypassing approval, sacrificing message quality for engagement
  6. Save and hand off:
    • Save the two-section report and structured experiment backlog under the Audience Growth Engine.
    • Link every recommendation to its source finding or explicit hypothesis.
    • Report exact artifact paths, next review date, approvals required, and deliberate non-actions.
    • Anti-patterns: publishing directly, omitting the future review point, claiming optimization success before the next measurement cycle

Report Design

Analysis

Summarize scope, defensible cross-cutting patterns, platform-specific signals, counterexamples, uncertainties, and data gaps.

Optimization

List prioritized next-week actions, testable hypotheses, measurement plans, owners, review gates, and the next review date.

Anti-Patterns

  • Treating optimization guidance as proven fact.
  • Recommending actions that cannot be traced to evidence or a clearly labeled hypothesis.
  • Optimizing only for total engagement without considering reach, audience quality, mission, or measurement comparability.
  • Drafting or publishing content when the task is interpretation and planning.
  • Producing a plan too broad to execute or evaluate in one cycle.

Output

Produce a report with the two main sections Analysis and Optimization, plus a structured experiment backlog. Include source finding IDs, evidence strength, next-week actions, test definitions, approvals required, next review date, exact artifact paths, and a statement that no content was published.