Best credit card for online shopping — and when a flat-rate card beats a co-brand

The best credit card for online shopping is settled by a measurement you can make in twenty minutes, not by a ranking. Group three months of card statements by merchant. If more than roughly half your online spending lands on a single platform, a co-branded card for that platform will usually beat a generalist, because the platform is subsidising the reward and no ordinary card can match a subsidy. If it does not — and for most people it does not — a flat-rate cashback card wins, and it keeps winning when your shopping habits change.

Online retail is the one category where co-brands genuinely earn their keep. That is exactly why it is worth being precise about where the advantage ends.

Measure before you look at any card

Do this first. It decides everything that follows and takes less time than reading a comparison table.

Open the last three statements on every card you use, plus your UPI and net-banking history if you pay for anything online that way, and sort the online transactions by merchant. Then work out one number: the share of your total online spending that goes to your largest single platform.

Almost everyone overestimates it. The mental picture is “I buy everything on one app”. The statement usually shows two or three large platforms, a quick-commerce app that has quietly become the second-biggest line, food delivery, and a long tail of small direct-to-consumer sites that individually look trivial and collectively are not.

That distribution is the argument. A co-brand pays its accelerated rate on one column of that list and its base rate on everything else. A flat-rate card pays the same rate down the whole page. The more evenly your spending is spread, the more the base rate does the work — and the base rate on a co-brand is normally where its economics are recovered.

Compare the three months to each other as well. If your largest platform changes between them, you are not a co-brand customer, whatever the annual total says.

What a co-brand actually trades away

The trade is explicit once you name all three sides of it.

You get a rate on one platform that a generalist cannot match, because the retailer is paying for part of it to keep you buying there.

You give up the rate on everything else. Co-branded cards commonly pay a thin base rate off-platform, which is what funds the accelerated one.

And you accept that the reward is often not money. This is the part that gets skipped, and it is the largest hidden cost of the category.

There is a fourth cost that shows up only later: switching. A generalist card follows you when your shopping moves; a co-brand does not. Co-brands are renegotiated and retired regularly, and when one ends the accelerated rate goes with it while the card, and any fee attached to it, remains.

It also helps to know which of three shapes you are looking at, because “online shopping card” covers three different products.

The single-platform co-brand. Tied to one marketplace, accelerated there, ordinary elsewhere. SBI Flipkart and Amazon Pay ICICI are the two best-known examples in India. This is the shape that needs real concentration to justify.

The ecosystem co-brand. Tied to a group of brands rather than one storefront, paying its reward in the group’s own currency. Tata Neu HDFC Bank is the clearest example. Broader coverage than a single platform, but the currency is correspondingly more locked in.

The multi-merchant online card. Not a co-brand at all — accelerated earning across a list of online merchants the issuer maintains, with a base rate everywhere else. HDFC Millennia sits here. Because the issuer can revise that merchant list, check the current one rather than the one you remember.

The platform-currency problem

A reward paid in a platform’s own coins, vouchers or store credit is not equivalent to cashback, and comparing the two at face value overstates the co-brand every time.

Convert to rupees first. Find out what one unit redeems for on the route you will actually use, not the best route in the brochure. Then apply the discounts:

• Expiry. Platform currencies routinely lapse. Anything you do not spend inside the window is worth zero, and the breakage is priced into the programme.
• Minimum redemption. If a balance below a threshold cannot be spent at all, small earnings are dead until they accumulate.
• Partial-payment caps. Many platform currencies can only pay for a limited share of an order, so you cannot clear a balance in one purchase even when you want to.
• Category restrictions. The currency frequently cannot be used on the very categories where the platform’s own margin is thin — electronics, groceries, gold.
• Single-retailer risk. The whole balance is contingent on you continuing to shop there.

Value a restricted platform currency at a meaningful discount to its stated rupee value. The exact haircut depends on the restrictions above, but treating it as worth face value is the one choice that is definitely wrong. Direct cashback or a statement credit carries none of these conditions, which is why a lower cashback rate frequently beats a higher points rate once both are in rupees — our cashback card round-up works through that conversion.

Where the advertised rate stops

Every accelerated online rate stops somewhere, and the two places it stops are the monthly ceiling and the exclusion list.

The cap is a per-statement-cycle limit on how much accelerated reward you can earn. Past it, spending reverts to the base rate for the rest of the cycle. A cap that sits below your normal monthly platform spend converts the headline rate into a fixed small monthly amount, and at that point the card’s real return is the base rate plus a rounding error.

The exclusions are more varied but the recurring ones on marketplaces are consistent: gold and bullion, gift cards and vouchers, wallet top-ups, fuel, rent, insurance, government payments — and, importantly, any transaction converted to EMI. That last one catches people, because the large purchases most worth earning on are exactly the ones most likely to be put on instalments. If you convert, check whether you also gave up the reward, and run the EMI arithmetic on what the conversion actually costs before agreeing to it.

This page names no card’s rate, cap or fee on purpose. Our research pass found published credit card data to be the least reliable category we encountered anywhere on this site: cards described as lifetime free that in fact carry joining fees of several thousand rupees, and multipliers quoted without the earn base they multiply, which turned one premium-sounding rate into roughly half a percent effective. Read the Most Important Terms and Conditions on the issuer’s own website, for your exact variant — variants within a card family differ — and believe the issuer over any aggregator.

The sale-season distortion

Here is the effect that most “best card for online shopping” pages ignore, and it is often larger than the entire reward argument.

During the large seasonal sales, platforms and card issuers run instant-discount tie-ups: a percentage off the order value, applied at checkout, capped per transaction, and available only to cardholders of one named issuer. A single large purchase made under one of those offers can be worth more than a year of a co-brand’s base earning.

The strategic consequence is not the one you would expect. These tie-ups are struck with the biggest issuers, because that is where the cardholder volume is. So the card that captures the most sale-season value is not the most rewarding niche card — it is any card from a large issuer, held for that reason. If you shop the sales seriously, holding two no-fee cards across two large issuers covers more offers than optimising one clever card.

Two cautions. The discount usually carries a minimum order value, a rupee cap, and sometimes an EMI requirement — which may cost you the reward, as above. And a discount is only a saving on something you were going to buy anyway.

When a generalist card wins

For most people, most of the time. Specifically:

• Split spending. No platform reaches roughly half your online total.
• A long tail. A large share of your spending is at small merchants no co-brand covers.
• An unstable pattern. Your biggest platform changes between quarters.
• You will not manage multiple cards. One card used consistently beats three used approximately, and a card you forget to reach for at the right merchant is worth nothing.
• You want the reward in money. Cashback or a statement credit needs no redemption strategy and cannot expire.

A flat-rate card is also the more defensive choice, because it does not care which app wins the next five years. Start from the lifetime-free options: a card with no annual fee cannot become a bad deal, only a mildly useless one, whereas a fee-paying co-brand whose platform you drift away from is a recurring cost against a benefit you no longer earn.

The card in this category that pays us nothing

Amazon Pay ICICI belongs in any honest answer to this question, and it has no affiliate programme we can join. We earn nothing if you take it, and we are naming it anyway because for readers whose online spending concentrates on Amazon it is frequently the correct card — it pays into a spendable balance rather than into a points catalogue, which is a different kind of reward from a store credit good at one shop only. We have not verified its current rate, cap or fee against the issuer’s own terms and will not publish any of them here; read them on the issuer’s page before you apply, and treat any figure published elsewhere as unconfirmed. ICICI describes it as an invite-only programme and publishes no invitation criteria, so treat it as a card to accept if it is offered rather than one to apply for — see ICICI lifetime free credit cards.

It has the same structural weakness as every co-brand here: the accelerated rate applies to one platform, the rate elsewhere is ordinary, and the currency is still a platform currency even if a comparatively liquid one. Measure your concentration and read the current terms before assuming any of that still holds.

If we steered you to a card we get paid for while a card we do not get paid for is the better fit, nothing else on this site would be worth reading. Our advertiser disclosure says so; this page is where it costs us something.

Saved cards and tokenisation

Under the RBI’s card-on-file rules, no entity in the card transaction chain other than the card issuers and the card networks may store actual card data. The date is worth getting right, because most coverage still gives the wrong one: the original tokenisation circular set 1 January 2022, that deadline was extended twice, and the restriction finally took effect on 1 October 2022. A merchant may keep only the last four digits of the card number and the issuer’s name. What a shopping site saves when you tick “save this card” is therefore a token, not your card number.

Tokenisation is free to you, and is done only with your explicit consent through an additional factor of authentication. The token is unique to a combination of the card, the entity requesting it and the device, so it is not a portable copy of your credentials. Practically: allow tokenisation on sites you use often.

None of this protects you from the charges that actually cost money on a card, which are interest and late fees rather than fraud — the charges worth understanding are on the statement, not in the checkout flow. One revolved balance can easily cost more than a year of a shopping card’s rewards is worth.

How to decide

Run it in this order.

1. Measure. Three months of statements, grouped by merchant. Find your largest platform’s share of online spend, and check it is stable across all three months.
2. If one platform is above roughly half and stable, shortlist its co-brand. Then read that card’s Most Important Terms and Conditions for the cap, the exclusion list, and what the reward currency can and cannot buy.
3. Convert the reward to rupees, discounted for expiry, minimum redemption and category limits, then compare it against a flat cashback rate applied to your whole online total — not just to the platform.
4. If the answer is close, take the flat-rate card. It is worth less on paper and more in practice, because it needs no management and survives a change of habit.
5. Separately, hold a card from a large issuer for sale-season instant discounts, which may be worth more than the reward argument you just settled.

If your spending is spread rather than concentrated — and the statement, not the impression, decides that — the honest answer is that no shopping card is your best card, and you should be picking from the general round-up on base rate and fee instead.

Common questions

Is a co-branded shopping card better than a flat cashback card?

Only if your online spending is genuinely concentrated on that one platform. The platform subsidises the accelerated rate, which is why no generalist card can match it — but the subsidy applies to one merchant, and the same card typically pays a weak rate everywhere else. Add up three months of statements by merchant before deciding. If your largest platform is under about half of your online spending, the base rate on the remainder usually matters more than the accelerator, and a flat-rate card wins.

Why do you not list the reward rates for each shopping card?

Because we could not verify them against issuer documents. Our research pass found published card data wrong often enough to be dangerous — cards described as lifetime free that in fact carry joining fees of several thousand rupees, and a reward rate quoted as a multiplier without its earn base, which turned a premium-sounding rate into roughly half a percent. Read the Most Important Terms and Conditions on the issuer’s own site, for your exact card variant, and treat any figure a comparison site publishes as unverified until you have matched it there.

Are platform coins and vouchers worth the same as cashback?

No. Convert them to rupees before comparing, then discount for the restrictions. Platform currencies commonly expire, need a minimum balance before you can redeem, cap the share of a single order they can pay for, and exclude categories. A reward you can only spend at one retailer is worth less than cash, and worth nothing if you stop shopping there. Statement credit or direct cashback carries none of those conditions, so it is worth its face value.

What is usually excluded from accelerated online rewards?

Exclusion lists vary by card, but the recurring ones on marketplace purchases are gold and bullion, gift cards and vouchers, wallet top-ups, fuel, rent, insurance premiums, government payments, and any transaction converted to EMI. Accelerated rates also carry a monthly ceiling, after which spending drops to the base rate for the rest of the cycle. Both the cap and the exclusion list live in the Most Important Terms and Conditions rather than in the marketing, and both are where the advertised rate usually stops being true.

Are my saved cards on shopping sites stored as card numbers?

They should not be. Under the RBI’s card-on-file rules — the 1 January 2022 deadline in the original tokenisation circular was pushed back twice and finally bit on 1 October 2022 — no entity in the card transaction chain other than the card issuers and card networks may store actual card data; a merchant may retain only the last four digits and the issuer’s name. What a shopping site saves instead is a token. Tokenisation is free to you, is done only with your explicit consent through an additional factor of authentication, and the token is unique to a combination of card, the entity requesting it and the device.

Sources

Rates and rules on this page were read directly from the following sources on the dates shown. Figures change — if you are about to act on one, confirm it at the source.

1. Tokenisation – Card Transactions: Permitting Card-on-File Tokenisation (CoFT) ServicesReserve Bank of India · checked 18 August 2026
2. Restriction on Storage of Actual Card Data [i.e. Card-on-File (CoF)] — RBI/2022-2023/95Reserve Bank of India · checked 18 August 2026
3. FAQs — Tokenisation of card transactionsReserve Bank of India · checked 18 August 2026
4. RBI (Commercial Banks — Credit Cards and Debit Cards: Issuance and Conduct) Directions, 2025Reserve Bank of India · checked 22 August 2026