name: Qualitative Assessment description: Assess a company's economic moat, EBITA margin trajectory, and organic growth trajectory from analyst reports, earnings transcripts, or long-form articles. Produces structured qualitative outlook that feeds into the Financial Modeling skill's assumption generation.

Qualitative Assessment Skill

Purpose

This skill reads qualitative source documents (analyst reports, earnings call transcripts, press articles) and produces a structured assessment of three dimensions that directly feed into the DCF model assumptions:

1. Economic Moat — determines terminal growth rate and margin sustainability
2. EBITA Margin Outlook — adjusts Stage 2 margin assumptions (+/- pp)
3. Organic Growth Outlook — adjusts Stage 2 growth assumptions (+/- pp)

Without this skill, the Financial Modeling skill falls back to pure historical trend continuation, which misses forward-looking information.

Prerequisites

• The document must already be classified (Skill 1) — ticker, company name, and document type known
• The company's output_data/TICKER/ directory must already exist (i.e., at least one financial report has been processed)
• TICKER_metadata.md should exist with a Financial History table for context

Inputs

• Classified document in processing_data/ (analyst report, transcript, or article)
• output_data/TICKER/TICKER_metadata.md — for existing assessments to compare/harmonize

Outputs

• Updated TICKER_metadata.md with Qualitative Assessment section (new or harmonized with existing)

Step-by-Step Instructions

Step 1: Read the Source Document

Read the classified document from processing_data/. Identify:

• Source type: Analyst report, earnings transcript, press article, etc.
• Source date: Publication or earnings call date
• Author/firm: e.g., "Morningstar (Dan Romanoff, CPA)"

Also read the company's TICKER_metadata.md to check for any existing Qualitative Assessment.

Step 2: Assess Economic Moat

Determine the company's economic moat rating: Wide, Narrow, or None.

For each moat rating, provide:

• Rating: Wide / Narrow / None
• Confidence: High / Medium / Low
• Source: Document name and date
• Three bullets of rationale — each should be specific, cite evidence from the document, and name the moat source (network effects, switching costs, intangible assets, cost advantage, or efficient scale)

💡 If the source document explicitly states a moat rating (e.g., Morningstar reports), use that rating and cite it. Your rationale should support or contextualize their assessment.

Step 3: Assess EBITA Margin Trajectory

Determine whether the company's EBITA margin will expand or shrink over the next 5 years, and by how much.

For the three bullets of rationale, consider:

• Revenue mix shifts (high-margin vs. low-margin segments growing at different rates)
• Operating leverage (fixed costs being spread over growing revenue)
• Investment cycles (R&D, capex ramps that pressure near-term margins)
• Competitive pricing pressure
• Regulatory costs

⚠️ Be conservative. A +2 pp expansion over 5 years is significant. Moves larger than ±4 pp require very strong evidence.

Step 4: Assess Organic Growth Trajectory

Determine whether the company's organic revenue growth will increase or decrease relative to the current L4Q trend, and by how much.

For the three bullets of rationale, consider:

• TAM expansion or saturation
• New product launches and adoption curves
• Competitive dynamics and market share trends
• Macro/cyclical headwinds or tailwinds
• Management guidance and forward-looking commentary
• Base effects (easy or tough YoY comparisons)

Step 5: Harmonize with Existing Assessment

If TICKER_metadata.md already has a Qualitative Assessment section:

1. Compare each dimension (moat, margin, growth) with the existing assessment
2. If the new source agrees with the existing assessment: keep the existing, update the "Last updated" date and source
3. If the new source disagrees:
   • Consider recency (more recent = more weight)
   • Consider source quality (Morningstar analyst > news article)
   • Consider confidence levels
   • Update the assessment if the new evidence is stronger; otherwise keep existing and note the disagreement in a comment

Step 6: Write Output to Markdown

Write or update the following section in TICKER_metadata.md:

---

## Qualitative Assessment

_Last updated: {date} | Source: {source_description}_

### Economic Moat

| Field      | Value                  |
| ---------- | ---------------------- |
| Rating     | **{Wide/Narrow/None}** |
| Confidence | {High/Medium/Low}      |
| Source     | {source and date}      |

1. **{Moat Source 1}**: {specific evidence from the document}
2. **{Moat Source 2}**: {specific evidence}
3. **{Moat Source 3}**: {specific evidence}

### EBITA Margin Outlook

| Field      | Value                      |
| ---------- | -------------------------- |
| Direction  | **{Expand/Shrink/Stable}** |
| Magnitude  | {+/- N pp over 5 years}    |
| Confidence | {High/Medium/Low}          |

1. {Specific rationale with evidence}
2. {Specific rationale with evidence}
3. {Specific rationale with evidence}

### Organic Growth Outlook

| Field      | Value                          |
| ---------- | ------------------------------ |
| Direction  | **{Increase/Decrease/Stable}** |
| Magnitude  | {+/- N pp over 5 years}        |
| Confidence | {High/Medium/Low}              |

1. {Specific rationale with evidence}
2. {Specific rationale with evidence}
3. {Specific rationale with evidence}

Step 7: Move Source Document

Move the processed document from processing_data/ to output_data/TICKER/:

• Rename to remove the _temp suffix
• Both the PDF and the markdown file

How Downstream Skills Use This Output

The Financial Modeling assumptions skill (skills/financial_modeling/assumptions/) reads this section to adjust historical trends:

If no qualitative assessment exists, the assumptions skill defaults to Narrow moat behavior and pure historical continuation.

Important Notes

• Be specific, not generic. Every rationale bullet should cite evidence from the source document — numbers, quotes, product names, dates. Avoid vague statements like "strong brand" or "good management."
• Three bullets per dimension, no more. Forces prioritization of the most important factors.
• Magnitude matters. "+2 pp" and "+4 pp" produce very different DCF outcomes. Be precise and conservative.
• Confidence is about the evidence quality, not your personal certainty. A Morningstar deep-dive = High. A single news article = Low.
• The margin and growth outlooks are relative to the current L4Q trend, not absolute targets. If L4Q margin is 38% and you say "Expand +2 pp", the assumptions skill will set Stage 2 margin to 40%.

Reference

Based on extract_qualitative_assessment() in tiger-cafe/app/app_agents/qualitative_extractor.py, adapted from LLM-only assessment to document-grounded analysis.