Analytics
Track platform usage, document processing metrics, and quality insights.
Overview
The Analytics dashboard provides insights into:
- Usage - Documents processed, pages, credits
- Quality - Extraction accuracy, chat confidence
- Performance - Processing times, success rates
- Cost - Credit consumption, trends
Usage Metrics
Documents Processed
Track document processing volume:
- Total documents
- By file type (PDF, image, spreadsheet)
- By status (completed, failed)
- Over time (daily, weekly, monthly)
Pages Processed
Monitor page throughput:
- Total pages
- Average pages per document
- Processing rate (pages/minute)
Credits Used
Track credit consumption:
- Total credits used
- By operation type
- Remaining balance
- Usage trends
Quality Metrics
Extraction Accuracy
Measure extraction quality:
- Overall confidence - Average confidence across extractions
- Field-level confidence - Confidence by field type
- Ground truth comparison - Accuracy vs verified data
- Correction rate - How often extractions need manual corrections
Chat Confidence
Monitor chat response quality:
- Response confidence - Average confidence score
- Citation accuracy - How well citations match content
- User feedback - Thumbs up/down ratings
- Session length - Average questions per session
Performance Metrics
Processing Time
Track processing speed:
| Metric | Description |
|---|---|
| Average parse time | Time to parse documents |
| Average extraction time | Time to run extractions |
| Average chat response time | Time to generate responses |
| P95 latency | 95th percentile response time |
Success Rate
Monitor reliability:
- Processing success rate
- Extraction success rate
- Queue depth
- Retry rate
Dashboard Views
Summary View
High-level overview of key metrics:
- Documents processed (current period)
- Credits used vs allocation
- Success rate
- Active knowledge bases
Trends View
Time-series charts showing:
- Daily/weekly/monthly volumes
- Period-over-period comparisons
- Trend lines and forecasts
Drilldown View
Detailed breakdowns by:
- File type
- Extraction schema
- Knowledge base
- Time period
Filtering
Filter analytics by:
| Filter | Options |
|---|---|
| Date range | Last 7/30/90 days, custom |
| Knowledge base | Specific KB or all |
| Document type | PDF, image, etc. |
| Status | Completed, failed |
Export
Export analytics data:
Formats
| Format | Description |
|---|---|
| CSV | Spreadsheet-compatible |
| JSON | Raw data |
| Formatted report |
Export via API
curl -X GET "/api/analytics/export?format=csv&period=30d" \
-H "Authorization: Bearer YOUR_API_KEY" \
-o analytics.csvScheduled Reports
Automate report delivery:
Create a Schedule
- Go to Analytics → Reports
- Click Schedule Report
- Configure:
- Report type (usage, quality, cost)
- Frequency (daily, weekly, monthly)
- Recipients (email addresses)
- Format (PDF, CSV)
Report Types
| Type | Contents |
|---|---|
| Usage Report | Documents, pages, credits |
| Quality Report | Confidence, accuracy, feedback |
| Cost Report | Credits by operation, projections |
| Executive Summary | High-level overview |
Real-time Monitoring
Monitor processing in real-time:
- Active jobs count
- Queue depth
- Processing rate
- Recent completions
Access via Analytics → Real-time or API:
curl -X GET "/api/analytics/realtime" \
-H "Authorization: Bearer YOUR_API_KEY"Knowledge Base Analytics
Per-knowledge-base insights:
- Query volume
- Popular queries
- Response times
- User feedback
- Document citations
Access from the knowledge base page under Analytics tab.
Custom Queries
Build custom analytics queries:
curl -X GET "/api/analytics/query" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"metrics": ["documents_processed", "credits_used"],
"dimensions": ["file_type"],
"period": "30d",
"aggregation": "daily"
}'Best Practices
- Review weekly - Check metrics regularly
- Set baselines - Establish normal ranges
- Monitor anomalies - Watch for unusual patterns
- Track quality - Don’t just count volume
- Export for analysis - Use external tools for deep dives
- Automate reports - Schedule reports for stakeholders
API Reference
See the Analytics API for programmatic access.
Last updated on