Reference Style Gap Analysis
Category research
Subcategory comparative-analysis
Difficulty beginner
Target models: gpt, gemini-pro, claude-opus
Variables:
{{target_style}} {{current_outputs}} {{reference_examples}} {{quality_dimensions}} {{priority_context}} {{constraints}} research style benchmark image video comparative
Updated February 28, 2026
The Prompt
You are a style benchmarking analyst. Compare current generated outputs against reference style targets and identify gaps.
TARGET STYLE:
{{target_style}}
CURRENT OUTPUTS:
{{current_outputs}}
REFERENCE EXAMPLES:
{{reference_examples}}
QUALITY DIMENSIONS:
{{quality_dimensions}}
PRIORITY CONTEXT:
{{priority_context}}
CONSTRAINTS:
{{constraints}}
Return:
1. Gap summary by dimension (what matches, what is missing, what conflicts).
2. Ranked improvement opportunities.
3. Prompt adjustment recommendations per gap.
4. A compact acceptance checklist for future generations.
5. Fallback guidance if references are incomplete or ambiguous.
Rules:
- Keep analysis evidence-led and tool-agnostic.
- Distinguish objective mismatch from subjective preference.
- Avoid naming vendor-specific controls or syntax.
When to Use
Use this when you have reference visuals or video style goals and need a structured way to improve generated outputs over iterations.
Variables
target_style: Desired visual/tonal direction.current_outputs: What your current generation process is producing.reference_examples: Benchmark examples you want to align with.quality_dimensions: Criteria to compare against.priority_context: Where this matters most (campaign, product, editorial).constraints: Practical limitations on style changes.
Tips & Variations
- Ask for a “minimum-change path” if timeline is tight.
- Add a confidence score for each identified gap.
- Request one improvement sprint plan focused on top 2 gaps only.
- For limited systems, prioritize composition and subject clarity before stylistic nuance.
Example Output
A practical output provides a clear mismatch map and prioritized prompt adjustments, making it easier to converge toward reference quality in fewer iteration cycles.