claude-code-skills-reference/ppt-creator/references/RUBRIC.md

PPT Quality Scoring Rubric

Purpose: Systematically evaluate presentation quality and identify areas for improvement. A score ≥ 75/100 is required before delivery. If score < 75, refine the weakest items and re-score (max 2 iterations).

---

Scoring System

• Total Score: 100 points (10 items × 10 points each)
• Passing Threshold: ≥ 75 points
• Rating Scale (per item):
  • 9-10: Excellent (exceeds expectations)
  • 7-8: Good (meets expectations)
  • 5-6: Acceptable (minor improvements needed)
  • 3-4: Weak (significant improvements required)
  • 0-2: Poor (fundamental issues, must fix)

---

1. Goal Clarity (0-10 points)

What: Are the audience, objective, and call-to-action (CTA) clearly defined and documented?

Scoring Criteria:

• 10: Audience, objective, and CTA explicitly stated and tailored; assumptions documented
• 8: Audience and objective clear; CTA present but could be more specific
• 6: Audience/objective vague; CTA generic (e.g., "let's discuss")
• 4: Missing audience definition or objective; no clear CTA
• 2: Presentation lacks clear purpose or intended action

How to Check:

• Review INTAKE.md responses and archive.txt
• Check final slide for specific CTA (not "Thank you" or "Questions?")
• Verify speaker notes mention audience and goal

Example Scores:

• 10: "After this 15-minute presentation, coffee enthusiasts will try at least one new brewing technique within the next week."
• 6: "This presentation is about coffee brewing for people interested in coffee."
• 2: "Talk about coffee."

---

2. Story Structure (0-10 points)

What: Is the Pyramid Principle applied? (One conclusion → 3-5 first-level reasons → evidence)

Scoring Criteria:

• 10: Clear pyramid structure; conclusion upfront; logical flow from reasons to evidence
• 8: Pyramid structure present but hierarchy could be clearer
• 6: Some structure but not consistently pyramid-style (e.g., conclusion buried at end)
• 4: Scattered points without clear logical connection
• 2: No discernible structure; random order

How to Check:

• Review storyline in archive.txt or WORKFLOW Stage 2 output
• Verify cover slide states main conclusion
• Check that 3-5 body sections support the conclusion
• Ensure evidence supports reasons (not random facts)

Example Scores:

• 10: Cover: "Master three variables for great coffee" → Sections: Grind / Temp / Time → Each with 2-3 evidence slides
• 6: Conclusion at end; sections exist but don't clearly support a single main point
• 2: Slides jump between topics with no connective thread

---

3. Slide Assertions (0-10 points)

What: Are slide headings assertion sentences (testable claims), not topic labels?

Scoring Criteria:

• 10: All slide headings are complete, testable assertion sentences
• 8: Most headings are assertions; 1-2 topic labels remain
• 6: Mix of assertions and topic labels (50/50)
• 4: Mostly topic labels with few assertions
• 2: All headings are topic labels (e.g., "Revenue", "Background", "Methodology")

How to Check:

• Review slide titles in slides.md
• Test: Can you agree/disagree with the heading? (If yes → assertion; if no → topic label)
• ✅ Assertion: "Finer grind size extracts flavors faster"
• ❌ Topic label: "Grind Size"

Example Scores:

• 10: Every slide heading is a complete sentence making a claim
• 6: Half are assertions ("Revenue grew 35%") and half are topics ("Q3 Results")
• 2: All headings are one-word or topic-style ("Introduction", "Conclusion")

---

4. Evidence Quality (0-10 points)

What: Is evidence sufficient, credible, and properly cited?

Scoring Criteria:

• 10: All claims backed by data/examples/citations; sources cited; units/methodology clear
• 8: Most claims have evidence; minor gaps in citations or methodology
• 6: Some claims lack evidence; sources missing or vague
• 4: Many unsupported claims; no citations; unclear data provenance
• 2: Assertions without any evidence or support

How to Check:

• Verify each assertion slide has chart/table/example/case study
• Check footer for source citations (e.g., "Source: XYZ, 2024")
• Confirm data units, time ranges, and methodology are specified
• Look for placeholder charts with "Data required: [fields]" if data unavailable

Example Scores:

• 10: "68% report bad cup experiences (Source: NCA 2024 Survey, n=1,200 home brewers)"
• 6: "Most people have bad coffee sometimes" (no data, no source)
• 2: "Coffee is important" (pure opinion, no evidence)

---

5. Chart Fit (0-10 points)

What: Are charts correctly selected, labeled, and easy to read?

Scoring Criteria:

• 10: Chart type matches data/message (per VIS-GUIDE); axes, units, source, alt text all present
• 8: Chart type correct; minor labeling gaps (e.g., missing unit or source)
• 6: Chart type suboptimal (e.g., pie chart with 7 slices); some labels missing
• 4: Wrong chart type for data; poor labeling; hard to interpret
• 2: No charts, or charts are misleading/unreadable

How to Check:

• Review chart selection against VIS-GUIDE.md Chart Selection Dictionary
• Verify all charts have: axis labels, units, data source, alt text
• Check readability: Can you understand the chart in 5 seconds?

Example Scores:

• 10: Line chart for time series, properly labeled, source cited, alt text provided
• 6: Bar chart OK but Y-axis missing unit; no source citation
• 2: 3D exploded pie chart with 10 slices and no labels

---

6. Visual & Accessibility (0-10 points)

What: Does the design meet WCAG AA standards and STYLE-GUIDE specs?

Scoring Criteria:

• 10: Contrast ≥4.5:1 (text) / ≥3:1 (UI); font sizes ≥18pt body / ≥34pt heading; white space ≥40%; alt text present
• 8: Minor accessibility issues (e.g., one chart with 4.2:1 contrast)
• 6: Multiple contrast or font size issues; some alt text missing
• 4: Poor contrast (<3:1), tiny fonts (<14pt), cluttered layout
• 2: Unreadable (light gray on white, <12pt fonts, no alt text)

How to Check:

• Use WebAIM contrast checker on text/background combos
• Measure font sizes (headings ≥34pt, body ≥18pt)
• Estimate white space (aim for 40-50% empty)
• Verify alt text for all images/charts
• Check colorblind-friendliness (use simulator)

Example Scores:

• 10: Dark text (#1F2937) on white, 36pt headings, 20pt body, 45% white space, all alt text present
• 6: Some 16pt body text, one chart missing alt text, 25% white space (crowded)
• 2: Light gray text on white, 12pt font, no margins, no alt text

---

7. Coherence & Transitions (0-10 points)

What: Do slides flow logically with smooth chapter and page transitions?

Scoring Criteria:

• 10: Clear section dividers; speaker notes include transitions; logical progression
• 8: Good flow overall; minor abrupt jumps
• 6: Some disjointed transitions; missing section dividers
• 4: Slides feel disconnected; unclear how one leads to next
• 2: Random order; no transitions or connective tissue

How to Check:

• Review speaker notes for transition phrases (e.g., "Now that we've covered X, let's explore Y")
• Check for section divider slides between major chapters
• Verify table of contents matches actual slide sequence

Example Scores:

• 10: Section dividers present; every speaker note ends with "This leads us to [next topic]..."
• 6: Flow is OK but one abrupt jump from "Problem" to "Conclusion" skipping "Solution"
• 2: Slides seem shuffled; no clear reason for order

---

8. Speakability (0-10 points)

What: Are speaker notes natural, well-paced (45-60 sec/slide), and easy to deliver?

Scoring Criteria:

• 10: All notes 45-60 sec; natural spoken language; structured (opening → assertion → evidence → transition)
• 8: Most notes well-paced; minor awkward phrasing
• 6: Some notes too long (>90 sec) or too short (<30 sec); some written-style language
• 4: Many notes poorly paced; reads like an essay, not speech
• 2: No speaker notes, or notes are bullet-point lists (not full script)

How to Check:

• Read notes aloud and time them
• Listen for natural speech patterns (contractions, questions, pauses)
• Verify structure: opening hook → core assertion → evidence walkthrough → transition

Example Scores:

• 10: "Now, here's the key insight: finer grind means more surface area. Think of it like sugar—powdered sugar dissolves instantly, while sugar cubes take forever. [PAUSE] Let's see how this plays out across five grind sizes..."
• 6: "The slide shows five grind sizes ranging from espresso to cold brew. Each has different particle size." (too dry, too short)
• 2: "• Espresso grind • Pour-over grind • French press grind" (bullet list, not script)

---

9. Deliverables Complete (0-10 points)

What: Are all required output files present and correctly formatted?

Scoring Criteria:

• 10: All files present and correct: slides.md, notes.md, refs.md, assets/*.png (if applicable), README.md
• 8: All core files present; minor formatting issues or missing README
• 6: Missing one deliverable (e.g., refs.md) or major formatting issue
• 4: Missing multiple deliverables or files are incomplete
• 2: Only partial output (e.g., slides.md exists but no notes or charts)

How to Check:

• Verify /output/ directory contains:
  • slides.md (Markdown slides with YAML frontmatter, speaker notes)
  • notes.md (Full speaker script + assumptions section)
  • refs.md (Citations and sources)
  • assets/*.png (charts, if data was provided)
  • README.md (explains file structure)
  • presentation.pptx (optional, if python-pptx available)

Example Scores:

• 10: All 5-6 files present, properly formatted, no broken links
• 6: Missing refs.md or README.md; one broken chart image link
• 2: Only slides.md exists; everything else missing

---

10. Robustness (0-10 points)

What: Are gaps/assumptions documented, and fallback plans provided?

Scoring Criteria:

• 10: All assumptions documented in notes.md; placeholders for missing data include field lists; next steps clear
• 8: Most assumptions noted; minor gaps in fallback plans
• 6: Some assumptions undocumented; placeholder charts lack detail
• 4: Many assumptions hidden; no guidance for missing data
• 2: Assumptions concealed; no acknowledgment of limitations

How to Check:

• Review "Assumptions & Limitations" section in notes.md
• Check placeholder charts have "Data required: [field list]"
• Verify next steps or follow-up actions are mentioned (if applicable)

Example Scores:

• 10: "Assumptions: (1) Used default 15-min duration (user did not specify). (2) No data provided for extraction curves; placeholder included with required fields: temperature_f, extraction_pct, time_sec."
• 6: Assumptions partially noted but missing some; placeholders generic ("Add chart here")
• 2: No mention of assumptions; missing data silently ignored

---

Scoring Workflow

Step 1: Initial Scoring

1. Review the presentation against all 10 criteria
2. Assign 0-10 points for each item
3. Calculate total score (sum of 10 items)

Example Initial Scorecard:

1. Goal Clarity: 9/10 ✓
2. Story Structure: 8/10 ✓
3. Slide Assertions: 6/10 ⚠️
4. Evidence Quality: 7/10 ⚠️
5. Chart Fit: 8/10 ✓
6. Visual & Accessibility: 9/10 ✓
7. Coherence & Transitions: 7/10 ⚠️
8. Speakability: 8/10 ✓
9. Deliverables Complete: 9/10 ✓
10. Robustness: 8/10 ✓
────────────────────────
TOTAL: 79/100 ✓ (≥75, ready to deliver)

---

Step 2: If Score < 75, Identify Top 3 Weaknesses

1. Sort items by score (ascending)
2. Identify the 3 lowest-scoring items
3. Write specific improvement actions for each

Example (if total was 72/100):

TOP 3 WEAKNESSES:
1. Item 3 (Slide Assertions): Score 5/10
   - Problem: Slides 4, 7, 11 use topic labels ("Grind Size", "Temperature")
   - Action: Revise to assertion sentences:
     • Slide 4: "Finer grind size extracts flavors faster and more completely"
     • Slide 7: "Water between 195-205°F produces balanced, full-bodied coffee"
     • Slide 11: "Simple equipment upgrades ensure consistent results"

2. Item 4 (Evidence Quality): Score 6/10
   - Problem: Missing source citations on 3 charts; no methodology note
   - Action:
     • Add footer to charts: "Source: National Coffee Association, 2024"
     • Add methodology note in refs.md: "Survey n=1,200 home brewers, margin of error ±3%"

3. Item 7 (Coherence & Transitions): Score 6/10
   - Problem: Abrupt jump from Slide 8 (temperature) to Slide 9 (time); missing section divider
   - Action:
     • Insert transition slide: "Now that we've mastered grind and temperature, let's tackle the third variable: time"
     • Update speaker notes for Slide 8 to bridge: "...and this brings us to our final variable."

---

Step 3: Apply Improvements & Re-Score

1. Make the improvements
2. Re-score all 10 items
3. If new total ≥ 75 → deliver
4. If new total < 75 → repeat Step 2-3 (max 2 iterations total)

Example Re-Score:

1. Goal Clarity: 9/10 ✓
2. Story Structure: 8/10 ✓
3. Slide Assertions: 9/10 ✓ (improved from 5)
4. Evidence Quality: 8/10 ✓ (improved from 6)
5. Chart Fit: 8/10 ✓
6. Visual & Accessibility: 9/10 ✓
7. Coherence & Transitions: 8/10 ✓ (improved from 6)
8. Speakability: 8/10 ✓
9. Deliverables Complete: 9/10 ✓
10. Robustness: 8/10 ✓
────────────────────────
TOTAL: 84/100 ✓✓ (exceeds threshold, ready to deliver)

---

Iteration Limits

• Max Iterations: 2 rounds of improvements
• Why Limit: Avoid infinite refinement loop; deliver practical value quickly
• If Still < 75 After 2 Rounds:
  • Deliver with clear disclaimer: "This presentation scores [X]/100. The following items need further work: [list weakest 3 items]."
  • Provide improvement roadmap in notes.md

---

Common Score Ranges & Interpretations

Score Range	Interpretation	Typical Issues
90-100	Exceptional	Exceeds all criteria; publication-ready
75-89	Good (passing)	Minor polish needed; ready to present
60-74	Needs improvement	Missing some assertions, evidence, or accessibility fixes
45-59	Weak	Major structure or clarity issues; requires significant rework
0-44	Poor	Fundamental problems; restart from WORKFLOW Stage 2

---

Self-Evaluation Checklist (Quick Version)

Use this quick checklist before full scoring:

• Goal: Audience, objective, CTA documented and clear?
• Structure: Pyramid (conclusion → reasons → evidence)?
• Assertions: All headings are testable sentences?
• Evidence: All claims have data/examples/citations?
• Charts: Correct type, fully labeled, source cited?
• Accessibility: Contrast ≥4.5:1, fonts ≥18pt, alt text?
• Transitions: Smooth flow, section dividers, speaker notes?
• Speakability: Notes 45-60 sec, natural language?
• Deliverables: slides.md, notes.md, refs.md, assets/?
• Robustness: Assumptions documented, placeholders detailed?

If all checkboxes are ✓, score is likely ≥ 75.

---

Final Delivery Criteria

Before delivering to user, confirm:

1. Total score ≥ 75/100 (or 2 improvement iterations completed)
2. All deliverables in /output/ directory
3. Assumptions and limitations documented in notes.md
4. If score < 75, include improvement roadmap

Delivery Message Template:

✅ Presentation ready!

SCORE: [X]/100 (threshold: 75)
QUALITY: [Exceptional / Good / Needs improvement]

DELIVERABLES:
- /output/slides.md (Markdown deck, [N] slides)
- /output/notes.md (Speaker script + assumptions)
- /output/refs.md (Citations and sources)
- /output/assets/ ([N] charts)
- /output/presentation.pptx (if available)

NEXT STEPS:
- Review speaker notes and adjust for your personal style
- Replace placeholder charts with your data (use chartkit.py if needed)
- Customize colors/fonts per STYLE-GUIDE.md

[If score < 75: Add improvement roadmap here]

---

Next Steps: Once scoring is complete and score ≥ 75, proceed to Stage 8 (Package Deliverables) in WORKFLOW.md.