A Payment Gateway Is Not a Drop-In Component, and the Bill for Guessing Is a Migration

A Flutter developer writing on dev.to has published the criteria table and decision rule he arrived at after comparing Stripe, Razorpay and Cashfree, framed as which gateway an Indian app should actually integrate in 2026 ^[1]. It matters because he says the process cost him two failed integrations and one painful migration first, which is the real price of treating a gateway as a drop-in component ^[2][3].

The sequence is ordinary. In 2021 he integrated Stripe for a SaaS product whose users were mostly in the US, and describes it as smooth with a mature SDK ^[4]. A year later the same company launched an Indian consumer app and the CTO asked him to "just add Razorpay" ^[5]. His conclusion: the answer depends on where users pay, how they pay, and who holds the merchant account ^[6], and a US subscription service and a UPI-first Indian consumer base are not the same problem ^[7].

His criteria, in his weighted order, are India payment methods (UPI, cards, net banking), Flutter SDK quality, pricing and settlement, onboarding and compliance, verification and webhooks, and international support, which he singles out as the criterion that flips most recommendations ^[8][9].

The scores he publishes for the two gateways in the available text: Stripe gets 6/10 on India methods, since its core is cards and international payments ^[10], with UPI supported and improved since 2023 but not the product's center of gravity ^[11]; 9/10 on the Flutter SDK, where flutter_stripe is first-party and null-safe and PaymentSheet handles most of the flow ^[12]; and 6/10 on India pricing, with domestic MDR above Razorpay's and Cashfree's on equivalent volumes ^[13]. Razorpay gets 9/10 on India methods, with UPI, cards, net banking, wallets and EMI first-class and the UPI app-switching flow handled by the SDK, which he calls the hardest part of Indian payments to build well ^[14][15]; 8/10 on razorpay_flutter, with the trade-off that checkout is Razorpay's own UI, themed by colour rather than fully controlled ^[16][17]; and 8/10 on pricing, with settlement at T+1 to T+2 ^[18][19].

Added up, that is 21/30 for Stripe and 25/30 for Razorpay on the three rows scored ^[20], which is close enough to prove his point rather than settle it. The spread is entirely in where it lands: Razorpay leads India methods by three points and trails on SDK by one ^[21]. Since India methods is weighted first, a domestic consumer app resolves one way and a global one resolves the other. He is explicit that no gateway wins every row and that the choice is a function of the user base ^[22].

The money argument is the one to take seriously. At low average order value with UPI-heavy volume, he says the per-transaction difference compounds, and that he has watched clients pay noticeably more than they would have with an India-native gateway ^[23]. He also warns that the published rate card is not the number, because tiers and volume negotiation are the real story ^[24]. Against that, he says Stripe is the answer when revenue is global, when international cards are a large share, or when a marketplace needs Connect payouts, which he rates dramatically better than anything the Indian gateways offer ^[25][26].

Two things to check before you commit. Whether your checkout is branded enough that Razorpay's own UI becomes a constraint rather than a convenience ^[17], and whether international revenue is incidental or meaningful, because that is the flip ^[9]. Note also that the Cashfree scorecard is not present in the text available here, so treat the comparison as two-way for now ^[27].