The Future of MCCs: Dynamic Codes and AI-Powered Categorization

The Limitations of Today's 4-Digit System

The 4-digit MCC system has been the payment industry standard since the 1970s. It works, but it's crude:

Problem 1: Categories Are Too Broad

• MCC 5812 = "Eating Places & Restaurants"
• This includes fine dining ($200+ per person), food trucks ($8 sandwich), and everything in between
• A $500 Michelin-star dinner and a $5 street taco have the same MCC
• Risk scoring and rewards bonuses can't differentiate

Problem 2: Single Code Per Merchant

• A supermarket that serves food (MCC 5411) doesn't get credit for also operating a pharmacy (MCC 5912)
• An airport hotel (should be 3500) that also has a casino (should be 7995) gets coded as one or the other, not both
• A multi-purpose store with groceries, electronics, and clothing gets one MCC that misrepresents 80% of its business

Problem 3: Static Categorization

• Once assigned, an MCC rarely changes
• A restaurant that pivots to meal delivery during COVID stays coded as 5812 (dine-in), not 5499 (meal kit)
• A crypto startup that adds fiat on-ramps stays coded as 6051, even though they're now offering financial services

Problem 4: No Granularity in Merchant Details

• The MCC tells you the merchant type, not:
  • Price tier (fine dining vs. casual)
  • Cuisine type (Japanese vs. Italian)
  • Merchant size (mom-and-pop vs. chain)
  • Loyalty program participation
  • Sustainability rating (eco-friendly merchant)

Result: Rewards algorithms and fraud scoring remain blunt instruments. Banks can't optimize rewards around true customer preferences. Fraud engines misclassify risk.

Emerging Innovations: Visa & Mastercard's Next Generation

1. Visa's Dynamic MCC & Merchant Tokenization

Visa is piloting merchant-specific tokenization, which moves beyond the 4-digit MCC:

How it works:

• Instead of a generic MCC, the merchant sends additional structured data at point of sale:
  • Merchant ID (unique ID for that specific location/business)
  • Sub-category code (e.g., "Japanese Fine Dining" vs. "Fast Casual Ramen")
  • Transaction metadata (online vs. in-store, delivery vs. dine-in)
  • Real-time classification (powered by merchant's recent transaction history)

Example: Upscale sushi restaurant in Tokyo

• Old system: MCC 5812 (generic restaurant)
• New system: Merchant ID XYZ123 + "Japanese Fine Dining" + "in-store" + 4.8-star rating (from Tabelog/Google)
• Cardholder's bank: "This is a high-end restaurant. Apply 5x bonus (vs. standard 3x for MCC 5812)"

Rollout timeline: Visa began pilot testing in 2024 with premium credit cards (AmEx Platinum, Chase Sapphire Reserve). Full rollout expected by 2027–2028.

2. Mastercard's AI-Driven Real-Time Categorization

Mastercard is developing machine learning models that reclassify merchants in real-time:

How it works:

• ML model ingests: merchant name, location, transaction amount, time of day, customer demographics, historical transaction patterns
• Model outputs: probability distribution across MCC categories
• If confidence score is low (ambiguous merchant), model requests additional data from merchant
• Categorization is not static—it updates as the model learns

Example: A London establishment called "The Queen's Kitchen"

• Could be: restaurant (5812), nightclub (7998), bar (5813), catering (5811)
• Old system: Acquirer guesses → assigns one MCC, potentially wrong
• New system: ML analyzes 100 historical transactions from this location → determines 70% are dinner-time, 20% are late-night bar service → dynamically categorizes as 60% 5812 + 20% 5813 for rewards/fraud purposes

Benefit: Cardholders get accurate rewards even when merchants are ambiguous. Fraud engines see nuanced merchant behavior.

Rollout: Mastercard began testing in 2023. Expected broader deployment by 2026–2027.

3. Granular Merchant Categories (GMC)

The payment industry is developing Granular Merchant Categories—a 6–8 digit code instead of 4 digits:

Current MCC: 5812 (Restaurant) Proposed GMC: 5812-JP-FD (Japanese Fine Dining)

Breakdown:

• 5812 = Restaurant (base category)
• JP = Cuisine (Japanese)
• FD = Format/Tier (Fine Dining)

Other examples:

• 3500-BT-BOU (Boutique Hotel, Beach Resort)
• 4011-LCC (Low-Cost Carrier Airline)
• 5499-OM (Online Meal Kit / E-commerce)

Benefit: Card programs can offer ultra-specific bonuses ("5x on Japanese fine dining but only 3x on casual ramen"). Fraud models can be more precise.

Timeline: Still in design phase (2025–2026); pilot testing expected 2027–2028.

4. Network-Side Merchant Classification (Not MCC-Dependent)

Some innovators propose moving away from merchant-assigned MCCs entirely:

New model:

• Card networks (Visa/Mastercard/Amex) classify merchants directly based on transaction data, not asking merchants to self-report
• No reliance on merchant accuracy
• Classification updates monthly based on transaction patterns
• Each cardholder sees optimized categorization tailored to their card's bonus structure

Example: A "Starbucks" transaction

• Visa sees: transaction at Starbucks GPS location, $6 charge, time 8am weekday
• Visa's ML classifies as: 90% "Coffee Shop" (MCC 5812 equivalent) + 10% "Grocery" (MCC 5411 equivalent)
• Issues two transaction records: one for rewards (uses "Coffee Shop"), one for fraud (uses both)

Benefit: Merchants can't miscategorize. Fraud and rewards are maximally accurate.

Timeline: R&D stage; likely not production until 2028–2030.

How This Changes Rewards & Fraud Detection (Next 3–5 Years)

Rewards Get Smarter

Today (2026):

• Card: "3x on dining (MCC 5812)"
• Result: Generic restaurants earn 3x, food trucks earn 3x, fine dining earns 3x

Tomorrow (2029):

• Card: "3x on casual dining, 5x on fine dining, 2x on fast casual"
• Result: Machine learns your merchant preferences; bonuses adjust per transaction

Example: You're a luxury traveler who frequents 5-star hotels and Michelin restaurants.

• Old cards: 3x hotel bonus (too generic)
• New cards: 3x on budget hotels, 5x on luxury hotels (AI-detected based on your booking history)

For comparison: You frequent chain restaurants and budget hotels.

• New card: 1x on luxury hotels (wasted bonus for you), 5x on chain restaurants (high value)

Fraud Detection Gets Granular

Today (2026):

• High-risk MCCs like 7995 (casinos) trigger auto-blocks
• Side effect: Legitimate casino charges are blocked frequently

Tomorrow (2029):

• System classifies: "Luxury resort casino in Singapore" (low fraud risk) vs. "Online poker site" (high fraud risk)
• Luxury resort casino: approved easily
• Online poker: additional verification needed
• Same MCC, different treatment

Benefit: Fewer false declines for legitimate international charges, while fraud prevention stays strong.

---

Challenges to Adoption

1. Legacy System Inertia

The payment industry has been using 4-digit MCCs for 50+ years. Changing it requires:

• Retraining 10 million+ merchants on new codes
• Updating 1000+ acquirer bank systems
• Reprogramming 500+ million credit cards
• Updating compliance/tax systems worldwide

Timeline reality: Even optimistic estimates expect 10+ years for full adoption.

2. Privacy & Data Concerns

Granular merchant classification requires more data sharing:

• What cuisine do you prefer? (might trigger targeted food delivery ads)
• How often do you visit casinos? (might affect credit scoring)
• Do you favor organic/eco-friendly stores? (might be sold to marketers)

Regulatory hurdle: GDPR, CCPA, and other privacy laws will slow rollout.

3. Merchant Resistance

Some merchants don't want granular categorization:

• A casino disguised as a "resort" might not want "casino" sub-category (higher fraud scrutiny)
• A budget hotel might not want "budget" label (might reduce premium traveler bookings)

4. Standards Competition

Visa, Mastercard, Amex, and Discover are developing competing systems. No unified standard yet.

---

Opportunities for Cardholders

What you should do now:

1. Stay informed: Track Visa's merchant tokenization and Mastercard's AI pilots (both announced publicly)
2. Advocate for change: If your card issuer hasn't adopted dynamic MCC classification, ask them when they will
3. Expect better rewards: By 2028–2030, your credit cards should offer 80%+ more specific bonus categories
4. Prepare for higher standards: Fraud detection will be more sophisticated; keep your merchants updated during trips

---

Key Takeaway

The 4-digit MCC system is outdated, but replacement is slow. By 2028–2030, expect dynamic merchant classification powered by AI, sub-category codes, and real-time categorization. This means better rewards accuracy, fewer fraud freezes, and more personalized bonuses. The transition will take a decade; legacy systems will persist alongside new ones. Cardholders who understand MCCs now will adapt faster to these innovations.

---

Your Next Step

Ask your card issuer: "Do you currently use dynamic merchant categorization or AI-based MCC classification? When will you roll this out?" Their answer reveals how forward-thinking they are. If they say "never heard of it," consider switching to a fintech card (Wise, Revolut, N26) that's actively testing these innovations. The first movers will offer the best rewards by 2028.