Merchant & Category Intelligence

Overview

The Merchant & Category Intelligence engine is the brain of Credit Card Co-Pilot. It transforms raw transaction data—whether from an SMS alert, a Gmail receipt, or a manual search—into actionable insights. By accurately identifying the merchant, the spending category, and the transaction channel (Online vs. Offline), the system ensures you always use the card that maximizes your rewards.

Smart Merchant Recognition

At the core of the system is a normalization engine that resolves messy transaction strings into clean merchant identities.

• Brand Normalization: The system identifies that ZOMATO*1234 and ZOMATO-PAY-RESTAURANT both refer to the same merchant: Zomato.
• Merchant Matrix: High-traffic merchants (like Amazon, Flipkart, or Uber) are mapped directly to specific reward rules, bypassing generic category logic to provide 100% accurate reward forecasts.
• Contextual Metadata: For every identified merchant, the system tracks historical data, including the typical channel (e.g., "Online") and category (e.g., "Dining"), to speed up future recommendations.

Category & Channel Mapping

Reward programs are rarely simple; a card might offer 5% on "Dining" but only if the transaction is "Online." Our intelligence engine handles this multi-dimensional logic automatically.

Spending Categories

Transactions are classified into standardized categories, including:

• Shopping: Online marketplaces and retail.
• Dining: Restaurants, cafes, and food delivery.
• Travel: Airlines, hotels, and ride-sharing.
• Utilities: Bill payments and recharges.

Channel Awareness

The system distinguishes between different ways of paying, as these often trigger different reward tiers:

• Online: Web and mobile app transactions.
• POS (Point of Sale): Physical card swipes or taps at retail outlets.
• UPI: Payments made via the UPI network (crucial for RuPay credit card logic).

Data Ingestion & Refinement

The intelligence engine continuously learns from your transaction history via the Parser Review system.

Automated Ingestion

The system connects to your Gmail or SMS (with permission) to automatically ingest transaction events.

• Parsed: Transactions successfully identified and categorized.
• Ambiguous: Transactions where the merchant or amount is unclear and requires a quick user confirmation.
• Failed: Raw data that couldn't be automatically processed, allowing for manual correction.

The Parser Review Interface

Users have full control over their data quality through the Parser Review screen. This interface allows you to:

1. Validate Ambiguous Entries: Quickly confirm "Ambiguous" transactions to keep your spend tracking accurate.
2. Edit Merchant Details: Manually override a merchant label or category if the auto-parser misses a nuance.
3. Correct Channels: Switch a transaction from "Online" to "Offline" to see how it affects your reward milestone progress.

// Example of the metadata tracked during a merchant review
{
  "reviewVersion": 1,
  "reviewLastAction": "accept",
  "merchantLabel": "Amazon India",
  "merchantKey": "amazon_in",
  "channel": "online",
  "parseStatus": "parsed"
}

Practical Use Cases

Real-Time Search

Before making a purchase, use the Merchant Search tool. By typing a brand name, the intelligence engine matches the merchant to your specific wallet.

• Scenario: You search for "Apple." The engine identifies the category as "Electronics/Shopping" and the channel as "Online," then calculates which of your cards offers the highest accelerated rewards or milestone bonuses for that specific spend.

Milestone Forecasting

Because the system knows which merchants belong to which categories, it can "nudge" you toward specific spending targets.

• Scenario: If you are ₹2,000 away from a "Travel" milestone, the engine will prioritize travel-heavy cards in its recommendations when it detects you are searching for airlines or hotels.

Post-Transaction Feedback

Immediately after a transaction is ingested via SMS or Email, the system derives the context to show you how many points you just earned.

• Logic: derivePostTransactionContext(events) scans the latest ingestion logs, extracts the merchant identity, and prepares a summary of the rewards earned versus the rewards potentially missed.