Cashfree Payments vs PayU India: Which is Better in 2026?

The short answer

Nine times out of ten, for a typical India founder on UPI plus cards—Shopify or a custom checkout—we’d take Cashfree Payments. Lower headline MDR. Payouts in the same house. Onboarding that doesn’t feel like a housing-loan file. PayU India is the one when you’re already enterprise-scale, you want LazyPay BNPL without roping in another vendor, or your bank RM has baked PayU into a blended rate you won’t match on a self-serve quote.

Where Cashfree Payments actually wins

We ran split traffic on UPI and domestic cards for months. Cashfree isn’t “magic.” It’s unit economics plus ops. The P&L tells you first; the brand halo comes later—₹62L GMV at 1.75% domestic versus 2% standard is ₹15,500 a month before GST on fees. Petty cash? Not at Series A burn. (You still pay GST on that MDR line. RBI tokenisation and card-on-file flows don’t kill the finance headache—they move it.)

• Payouts-heavy product (gig, lending disbursements, marketplace settlements): bank transfer, UPI, card, wallet rails; ₹5/transaction on payouts versus building a separate payout stack or living with PayU-shaped gaps. PayU, for us, is gateway-first. “Where’s my neobank layer?” gets awkward fast.
• Marketplaces / Easy Split: closing the books without a spreadsheet cult every month has a price. PayU can route smartly across banks; when we asked four founders last quarter, split-ledger ergonomics was Cashfree’s pitch.
• API + docs when your eng headcount is 1.2 people. Zero glamour. Fewer 2am threads in #payments-war-room.
• MDR negotiation headroom at volume: 1.75% domestic cards/UPI/NB against PayU’s 2% standard—you start closer to where you want to end.

Where it loses: if your checkout team hates “good enough” UI and wants the glossiest hosted page or the smoothest D2C polish, Cashfree’s dashboard (and sometimes merchant-facing chrome) can feel built by people who ship reliable pipes, not Instagram ads. Works for us. Not for everyone.

Where PayU India actually wins

PayU’s the uncle at every family wedding since 2011—Gurugram, fat bank ties. Boring to some. When the acquirer relationship is what saves you from a dumb BIN outage on sale night? Indispensable.

• Smart routing across acquiring banks at real volume—travel peaks, EdTech admission windows, BFSI spikes—where one choking MID costs more than an ugly dashboard.
• PayU Verify and fraud knobs when chargebacks aren’t a slide in a deck; they’re a weekly CFO sync.
• LazyPay BNPL without you becoming a BNPL integration PM. D2C brands that want EMI plus BNPL without two extra vendors get it.
• Enterprise procurement comfort: custom MDR, settlement T+1 to T+2 (working capital math when you’re floating inventory on rupee razor margins), paperwork that scares SMBs and calms compliance in bigger shops.

It loses when you’re indie SaaS and you care about developer experience and integrated payouts: PayU’s dashboard age and DX gap versus Razorpay/Cashfree shows up in sprint one. Not fatal. Just friction priced in engineer temper.

Pricing, in INR, no spin

What they advertise: Cashfree ~1.75% on domestic cards/UPI/netbanking; PayU ~2% standard, enterprise custom. Both: ₹0 setup in typical positioning; GST on fees (both mention GST invoices—do the paperwork once, or your CA will prepone your anxiety). Neither gave us one clean “all FX stories” number for international; treat cross-border as its own sheet. Anything invoiced in USD—mark [USD] in your contract. Gateway PDFs hide there.

Scenario A — mid-market D2C: ₹50,00,000 GMV/month, blended domestic, full headline rate (you’ll negotiate; apples-to-apples before the phones ring).

• Cashfree: 50,00,000 × 1.75% = ₹87,500 in MDR
• PayU: 50,00,000 × 2% = ₹1,00,000 in MDR
• Gap: ₹12,500/month (~₹1,50,000/year) before GST. Add 18% GST on the fee (export versus domestic supply can bite—check with your CA, not a blog; e-invoicing thresholds and GST credit rules still tripped teams in 2025–26).

Scenario B — heavier: ₹2,00,00,000 GMV/month (₹24 cr/year).

• Cashfree: ₹3,50,000/month MDR at 1.75%
• PayU: ₹4,00,000/month at 2%
• ₹50,000/month delta = ₹6,00,000/year pre-GST—a junior ops hire, or half a performance marketer, depending how you count.

Hidden / second-order costs (the real spreadsheet):

• Settlement timing: Cashfree sells instant settlements as a product line; PayU T+1 to T+2 is float you don’t have—if cost of capital is 12–14% annualised, a few days on ₹50L–₹1 cr rolling balances isn’t zero. (Quick sketch: ₹1 cr two days late isn’t “crore drama,” but twelve months later your treasury person still cares.)
• Payouts: Cashfree from ₹5/payout at 2 lakh driver transfers adds up; PayU without integrated payouts/neobank layer means another vendor or bank APIs—integration and reconciliation time is rupees even when it’s off the MDR line.
• Add-ons: fraud stacks, premium routing, international optimisation—enterprise custom MDR on PayU can flip the headline story; Cashfree “lower at scale” only counts if your FM negotiates instead of pricing-page idolatry.

What we’d actually use each for

Case 1: twelve-person D2C on Shopify, ₹40L–₹60L MRR-ish GMV, heavy UPI + cards, influencer payouts, micro-marketplace of resellers → Cashfree (checkout plugins exist; payouts stay in-house; you trade dashboard sex appeal for fewer vendors).

Case 2: travel-aggregator-ish flows, peaky traffic, bank caps across multiple ACQs, compliance wants PayU Verify, CFO wants LazyPay for BNPL without a science project → PayU (routing and enterprise muscle beat “nice dashboard” when uptime is revenue).

Case 3: NBFC / lending—disbursements, collections choreography, token vault / card-on-file hygiene post-RBI norms, you need virtual accounts / Auto Collect and boring, repeatable APIs → Cashfree usually fills the fintech-shaped hole; PayU does many rails, but “no integrated payouts/neobank product” bites when money movement is the product.

Indian fit (GST, UPI, IST, support)

GST: both charge GST on MDR; invoices from dashboards. The real question isn’t “which gateway”—it’s whether your HSN/SAC mapping is clean for ITC as e-invoicing norms and thresholds keep mid-market finance honest.

UPI: both are in; Cashfree gets founder chatter on UPI success rates in the wild (anecdotal, plus our ops—YMMV by PSP and bank day). PayU leans on bank breadth + routing; peak behaviour follows merchant profile, not Twitter.

IST hours: Cashfree Mon–Sat 10am–7pm IST versus PayU Mon–Fri 10am–7pm IST. Saturday-heavy sale? Indian consumer default. That Saturday window can matter when checkout 502s mid-flash drop.

Foreigner test: neither is Stripe with “Mumbai” pasted on a US invoice (both INR-native here). International and FX still mean fine print like TRAI decode; if it bills [USD], catch it in procurement before someone signs on caffeine.

Migration: what’ll bite you

Cashfree → PayU:

• Plugins and theme glue (Shopify/Woo/Magento): redirects, webhooks, sometimes customer flows—plan a maintenance window, not lunch-hour toggle.
• Subscriptions / eMandate / UPI Autopay: mandates and tokens don’t teleport; customer comms on re-consent edges (RBI tokenisation—legally spicy, not only annoying).
• Token vault / saved cards: PayU OneTap shines once live; migrating vaulted PAN/token references is what PMs underestimate—export limits and PCI walls slow you.
• Payouts: Easy Split or payout IDs in your ledger? PayU-side rebuild means new reconciliation keys in the DW.

PayU → Cashfree:

• Bank routing rules tuned for PayU’s smart routing won’t copy-paste; decline patterns re-learn for a week—budget support load.
• LazyPay / BNPL: you replace BNPL distribution or drop a checkout lane—D2C feels that in conversion, not Jira.
• Settlement rhythm: instant settlement can break treasury models built for T+1/T+2 float—tell finance before eng pops the champagne.

Contracts: lock-in, chargeback handling fees, minimums in footnotes—gateways love the quiet page-7 rupee leak.

What we’d pick

Default Cashfree for Indian product companies that need money in, money out, fewer existential spreadsheet hours, especially when ₹10L–₹50L/year MDR delta is real on your GMV. PayU when you’re enterprise-shaped, routing-sensitive, or “BNPL-lazy” in the good way—LazyPay already in place—and you can stomach heavier onboarding.

Rough dashboard versus rough bank route on a bad Saturday—which one actually keeps you up, once you strip the procurement memo?

Things people actually ask

“Is Cashfree really cheaper if I do ₹2 cr per month?”
Published 1.75% vs 2% → ₹35L vs ₹40L MDR on ₹2 cr—₹5L/month (~₹60L/year pre-GST) before you negotiate. Enterprise PayU can shave it; Cashfree “lower at scale” can too—quote both on your real mix.

“Do I need to redo my GST invoice template?”
You need correct vendor GSTIN + fee HSN/SAC mapping on gateway invoices; switching changes who issues the fee invoice, not the law. Sync with CA before quarter close.

“PayU OneTap vs Cashfree token vault—who wins RBI tokenisation drama?”
Both can run card-on-file if wired right; migration of existing tokens hurts. Plan customer re-permission for edge cases, not demo happy path.

“Settlement T+2—is that stealing my working capital?”
Sometimes, in rupee terms—model two-day float on average daily balance; stack against instant settlement fees if any. Treasury decides; not marketing PDFs.

“UPI Lite / small-ticket flows—does either care?”
Both do UPI; success rates and issuer quirks vary by segment. Run a controlled A/ test on your traffic, not a Twitter scoreboard.

“We’re on PayU for LazyPay only—can we stay?”
You can. Two gateways or feature lock-in. If BNPL is 5–8% of GMV with high AOV, integration tax might still pencil—math it.

“Support on Saturday?”
Cashfree lists Sat in the IST window; PayU is Mon–Fri. Weekend-heavy business: not trivia.

“International—are we going to get [USD] invoices and ruin our forex policy?”
Read the MSA; some modules bill oddly. Flag [USD] in vendor master before audit season—reality rarely matches the brochure line on cross-border.

“Migration webhooks—will our ledger burn?”
Payload shapes differ; replay in staging, rebuild idempotency keys, budget one week of edge-case triage even when the plan says three days—because it always says three days.