Pipeline Overview
Decisio's data pipeline is designed for reliability, speed, and security. Data flows through six stages, each optimized for its specific purpose.
Core Concepts
Data Pipeline
Understanding how data flows through Decisio — from ingestion to decision execution and continuous learning.
Step 1
Data Ingestion
Connect your data sources and import historical data
- One-click integrations (Shopify, Zoho, etc.)
- CSV/Excel file uploads
- REST API for custom sources
- Automatic schema detection
Step 2
Data Processing
Clean, normalize, and enrich your data
- Missing value handling
- Outlier detection and treatment
- Data type normalization
- Duplicate detection
Step 3
Data Storage
Secure, tenant-isolated data warehouse
- Time-series optimized storage
- Full audit trail
- Point-in-time recovery
- Encryption at rest
Step 4
AI Analysis
Agents analyze data and generate decisions
- Pattern recognition
- Trend analysis
- Anomaly detection
- Prediction generation
Step 5
Decision Output
Recommendations with explanations and confidence
- Actionable recommendations
- Impact predictions
- Confidence scores
- Full reasoning trace
Step 6
Sync & Learn
Execute decisions and learn from outcomes
- Bi-directional sync
- Outcome tracking
- Model retraining
- Feedback loops
Data Types We Process
Products
Product master data including SKU, name, category, brand, price, cost, and attributes. This forms the foundation for all agent analysis.
Sales History
Transaction-level sales data with date, SKU, quantity, revenue, and channel. Used for demand forecasting and elasticity modeling.
Inventory
Stock levels by SKU and location, including warehouse, on-hand, reserved, and in-transit quantities.
Competitors
Competitor pricing data for price intelligence. Can be imported manually or through integrations with price tracking services.
Suppliers
Supplier information including lead times, minimum order quantities, and reliability scores for inventory optimization.
Real-time vs Batch Processing
Decisio uses a hybrid approach:
- Real-time: Critical events like stockout alerts, price changes, and urgent decisions
- Near real-time: Dashboard updates, sync status (every 5 minutes)
- Batch: Full model retraining, historical analysis (daily/weekly)
Data Quality
Data quality directly impacts AI accuracy. Decisio includes built-in data quality monitoring:
- Completeness Score: Percentage of required fields populated
- Freshness Score: How recently data was updated
- Consistency Score: Alignment across related data points
- Accuracy Indicators: Flags for potential errors or anomalies
You'll see data quality scores in your dashboard and receive alerts if quality drops below thresholds.
Data Retention
By default, Decisio retains:
- Transactional data: 3 years
- Aggregated analytics: Indefinite
- Decision history: 2 years
- Audit logs: 1 year
Enterprise plans include custom retention policies and data archival options.