Summary
Transaction categorization in Nance is highly accurate and improves over time, especially as you review and correct transactions.
How This Works in Nance
Nance uses a combination of AI, rule-based logic, and user feedback to categorize transactions as accurately as possible.
1. Strong Base Accuracy
- Categorization is based on:
- Merchant names
- Transaction descriptions
- Known patterns from similar users
- Most common transactions (e.g., food, fuel, travel) are categorized correctly by default
2. Learns from Your Behavior
- When you edit:
- Category
- Subcategory
- Contact
- Nance treats this as a learning signal
- Repeated corrections increase future accuracy for similar transactions
3. Confidence-Based Automation
- Nance only auto-applies categories when confidence is high
- In uncertain cases, transactions may be placed in Miscellaneous instead of guessing incorrectly
4. Context from Contacts (Including UPI)
- Contact recognition (especially from UPI transactions) adds extra context
- Helps improve both:
- Categorization accuracy
- Consistency across similar transactions
What Affects Accuracy
- Clear vs unclear merchant descriptions
- First-time or uncommon transactions
- Whether you correct and guide the system
Important to Know
- You can always edit categories anytime
- Accuracy improves the more you use Nance
- Consistent edits help refine the system further
Smart from day one.
Personal by day ten.