Call Transcript Analyzer CLI

A Node.js command-line tool that fetches call transcripts from Close CRM, analyzes them using OpenAI ChatGPT batch processing, and generates summary reports with percentage breakdowns.

Features

• Fetch Transcripts: Download call transcripts from Close CRM with flexible filtering options
• AI-Powered Analysis: Process transcripts using OpenAI ChatGPT with custom analysis questions
• Summary Reports: Generate reports with percentage breakdowns and visualizations
• Flexible Filtering: Filter by date range, SDR name, and deal stage
• Multiple Export Formats: Console output, JSON, and CSV exports

Prerequisites

• Node.js: Version 18.x or higher
• Close CRM API Key: Required for fetching transcripts
• OpenAI API Key: Required for analyzing transcripts

Installation

npm install -g call-transcript-analyzer

Or for local development:

git clone <repository-url>
cd call-transcript-analyzer
npm install
npm link

Configuration

1. Environment Variables

Create a .env file in your project directory:

# Required
CLOSE_API_KEY=api_your_close_crm_key_here
OPENAI_API_KEY=sk-your_openai_key_here

# Optional
OPENAI_MODEL=gpt-4o-mini
LOG_LEVEL=info
OUTPUT_DIR=./analysis-results

Available OpenAI Models:

• gpt-4o-mini (default) - Fast and cost-effective
• gpt-4o - More capable, higher cost
• gpt-4-turbo - Previous generation, good balance
• gpt-3.5-turbo - Fastest, lowest cost

Choose based on your accuracy requirements and budget.

2. Questions Configuration

Create a questions.json file defining your analysis questions:

[
  {
    "id": "q1",
    "text": "Did the SDR build rapport with the prospect?",
    "answerFormat": "multiple_choice",
    "options": ["Yes", "No", "Partially"],
    "includeInReport": true
  },
  {
    "id": "q2",
    "text": "Rate the SDR's objection handling (1-5)",
    "answerFormat": "rating_scale",
    "scaleRange": { "min": 1, "max": 5 },
    "includeInReport": true
  }
]

See config/questions.example.json for more examples.

Quick Start

1. Fetch Transcripts

call-analyzer fetch \
  --start-date 2025-10-01 \
  --end-date 2025-10-31 \
  --output ./analysis-results

2. Analyze Transcripts

call-analyzer analyze \
  --input ./analysis-results/transcripts \
  --questions ./questions.json \
  --output ./analysis-results/analyzed

3. Generate Report

call-analyzer report \
  --input ./analysis-results/analyzed \
  --questions ./questions.json

Commands

fetch - Download Call Transcripts

Fetch call transcripts from Close CRM.

call-analyzer fetch [options]

Options:

• --start-date <date> - Start date in YYYY-MM-DD format
• --end-date <date> - End date in YYYY-MM-DD format
• --sdr <name> - Filter by SDR name
• --stage <stage> - Filter by deal stage (Discovery, Demo, Negotiation, etc.)
• -o, --output <dir> - Output directory (default: ./analysis-results)

Example:

call-analyzer fetch --start-date 2025-10-01 --end-date 2025-10-31 --sdr "John Smith"

analyze - Analyze Transcripts with AI

Process transcripts using OpenAI ChatGPT batch API.

call-analyzer analyze [options]

Options:

• -i, --input <dir> - Input directory with transcript files
• -q, --questions <file> - Questions configuration file
• -o, --output <dir> - Output directory for analysis results
• --overwrite - Overwrite existing analysis results

Example:

call-analyzer analyze --input ./transcripts --questions ./questions.json --output ./analyzed

report - Generate Summary Reports

Generate aggregate reports from analyzed transcripts.

call-analyzer report [options]

Options:

• -i, --input <dir> - Input directory with analysis results
• -q, --questions <file> - Questions configuration file
• -f, --format <format> - Output format: console, json, or csv (default: console)
• -o, --output <file> - Output file path (required for json/csv)
• --start-date <date> - Filter results by start date
• --end-date <date> - Filter results by end date
• --sdr <name> - Filter results by SDR name
• --stage <stage> - Filter results by deal stage

Examples:

# Console output
call-analyzer report --input ./analyzed --questions ./questions.json

# Export to JSON
call-analyzer report --input ./analyzed --questions ./questions.json --format json --output report.json

# Export to CSV
call-analyzer report --input ./analyzed --questions ./questions.json --format csv --output report.csv

# Filtered report
call-analyzer report --input ./analyzed --questions ./questions.json --sdr "Jane Doe"

Advanced Usage

Filtering Examples

Filter by SDR and Date Range:

call-analyzer fetch \
  --sdr "John Smith" \
  --start-date 2025-10-01 \
  --end-date 2025-10-31 \
  --output ./results

Filter by Deal Stage:

call-analyzer fetch \
  --stage Discovery \
  --start-date 2025-10-01 \
  --end-date 2025-10-31 \
  --output ./results

Complete Workflow

# 1. Fetch transcripts
call-analyzer fetch \
  --start-date 2025-10-01 \
  --end-date 2025-10-31 \
  --output ./analysis

# 2. Analyze with questions
call-analyzer analyze \
  --input ./analysis/transcripts \
  --questions ./questions.json \
  --output ./analysis/analyzed

# 3. Generate report
call-analyzer report \
  --input ./analysis/analyzed \
  --questions ./questions.json

Logging

Logs are automatically stored in ./logs/ directory with daily rotation:

• File: ./logs/call-analyzer-YYYY-MM-DD.log
• Retention: 30 days
• Format: JSON with timestamps, call IDs, and error details

Error Handling

The tool handles common errors gracefully:

• Invalid API Keys: Clear error messages with setup instructions
• Rate Limiting: Automatic prompts to continue, skip, or abort
• Partial Failures: Continues processing remaining items and logs errors
• Empty Results: Helpful messages when no data matches filters

Development

Build

npm run build

Run Tests

npm test

Lint

npm run lint
npm run lint:fix

Troubleshooting

No Transcripts Found

• Check date range is correct
• Verify SDR name spelling (case-sensitive)
• Ensure API key has proper permissions

OpenAI Batch Timeout

• Large batches (100+ calls) can take 30-60 minutes
• Split into smaller batches if needed
• Check OpenAI API usage and limits

Missing Configuration

• Ensure .env file exists with API keys
• Verify questions.json is properly formatted
• Check file paths are correct

License

ISC

Support

For issues and questions, please check:

• Configuration files (.env, questions.json)
• Log files in ./logs/ directory
• Close CRM and OpenAI API status