All posts
2 July 2026 6 min readWallet strategyRewardsAI agent

How to Pick the Best Card in Your Wallet for Every Purchase

Reward caps, category exclusions and merchant-classification quirks mean the 'best' card changes purchase-by-purchase. Here is a repeatable framework — and how the CardAgent wallet agent automates it.

Most Indian card users leave 30–50% of their potential rewards on the table not because they own the wrong cards, but because they swipe the wrong one at the wrong merchant. A ₹1,000 Swiggy order on a generic 1% card leaves ₹80 of reward value on a Swiggy HDFC. A ₹40,000 flight on a lounge-focused card leaves ₹1,600 on a travel-portal card.

The four-question framework

  1. What's the merchant category — as coded by the network, not as you'd describe it?
  2. Which of my cards has an active category multiplier for that MCC?
  3. How much of this month's category cap have I already used?
  4. Are there any active exclusions (rent, wallet loads, fuel above a threshold)?

Answering these correctly in the checkout queue is unrealistic. Most people default to the one card they know pays 'something', and lose the difference.

Merchant classification is the hidden variable

Zomato dine-out orders sometimes classify as 'Restaurants' (MCC 5812) and sometimes as 'Miscellaneous Retail'. BluSmart classifies as 'Transportation' rather than 'Taxi'. IRCTC classifies as 'Government'. These aren't edge cases — they routinely flip which card is optimal.

A simple monthly cadence

  1. On the 1st, check remaining category caps for each card in your wallet.
  2. Route recurring debits (utilities, subscriptions) to the card with the widest 'everywhere' earn rate — usually a RuPay-on-UPI card.
  3. Route discretionary spend to the highest-multiplier card for that merchant category.
  4. On the 25th, top up any card that's within ₹5,000 of a milestone voucher threshold.

Even without an agent, this cadence typically recovers 20–30% of foregone rewards. With the agent, it's automated at the point of purchase.