A comparison table is a piece of software. It takes inputs, applies a schema, and returns a verdict. And like any software, its most dangerous behavior is the failure that throws no error.
The table I have been staring at this week is titled "Krak vs. PayPal vs. Cash App โ Fees, Reach and Bank Withdrawals." Three names, three columns, one implicit conclusion. Krak, a crypto payment app, sits in the same row as two payment giants. The framing did work before a single fee was printed. Put a small challenger beside a trillion-dollar incumbent, and you have quietly promoted it.
But that is not the anomaly. The anomaly is the row that isn't there. Wise is missing. So is the only column that determines what you actually pay โ the FX spread.
Let me set the ledger straight before I audit it. These three are not the same class of machine. PayPal runs a closed-loop internal ledger wrapped around fiat bank rails. Cash App runs a similar loop with limited Bitcoin support. Krak settles on-chain โ almost certainly stablecoins โ then exits back into a bank account. That last step matters more than any marketing claim admits.
The "Reach" column is where the framing gets interesting. Krak and PayPal both support USโUK personal transfers. Cash App does not โ it requires US residency, no cross-border peer-to-peer. A casual reader files that as a product weakness. It isn't. Cash App's gap is a licensing decision, not an engineering one. Block holds US state money-transmitter licenses; it never built the bilateral licensed corridor into the UK. Reach differences in this table are regulatory differences wearing a technical costume. I have learned to distrust comparison pages that skip this distinction โ they are usually top-of-funnel content dressed as neutral review.
Here is where I stop reading headline fees and start tracing actual money. I did this on Compound V2 once, reconstructing the cToken rounding behavior: the advertised interest rate and the realized cost diverged by cents that compounded into real value. Payments work the same way. The number on the fee page is never the number you pay.
Decompose the cost into four layers. L1 โ the nominal transfer fee: PayPal's cross-border P2P sits near 4โ5%, Cash App charges $0 domestically, Krak's is undisclosed. L2 โ the FX spread: the gap between the real mid-market rate and the rate you're quoted, roughly 3โ4% on PayPal, invisible because it is baked into the exchange rate rather than billed as a line item. L3 โ the withdrawal fee: Cash App instant withdrawal about 0.5โ1.75%, PayPal instant about 1.75%. L4 โ the on-chain network fee: a stablecoin transfer usually under a dollar.
Now read the structure. L1 is loud and marketed. L2 is silent and is where the margin lives. This is the free-but-spread model: advertise a clean number, capture value in the exchange rate. Every remittance business on earth runs it, and the crypto app is not exempt. On-chain settlement genuinely cuts clearing cost โ 24/7, near-instant, sub-dollar. But the user still has to land in a bank account. That exit re-enters the traditional system, and the cost re-materializes exactly there. The money saved on the rail is often eaten at the withdrawal. That is the single sentence the table exists to avoid.
So "crypto payments are cheaper" holds only under strict conditions: both sides stay in stablecoin, and neither touches a bank. The moment "Bank Withdrawal" appears in the title โ and it does โ the advantage narrows to a spread you cannot see.
The licensing layer is the real moat, and it is the layer a fee comparison cannot render. Cross-border P2P requires compliance in both the sending and receiving jurisdiction. PayPal spent years building that bilateral authorization. Krak's UK-side status is undisclosed โ either it holds an FCA authorization itself, which takes years, or it rides a local partner, which introduces a single point of failure. When I traced FTX's hot wallets, the pattern that mattered was not the visible transactions but the missing segregation. Here the missing field is Krak's funding, reserve proof, and banking partners. Silence speaks louder than the proof.
Now the part that reads like a vulnerability, not a verdict. The comparison chose its opponents. PayPal is the most expensive option in the set and the most familiar. Cash App is the narrowest โ US-only. Beating the weakest two is not a market position; it is a content-marketing signature. Ghost in the audit: the absence of Wise tells you more than any fee row. Wise has already compressed the USโUK corridor to near mid-market with a transparent, explicit fee. If Krak cannot beat Wise on total cost or arrival time, the cheaper story fails on the most competitive, most transparent corridor in the world โ which is also the worst possible test bed for proving a new rail wins.
There is a deeper asymmetry. Krak is custodial. So are PayPal and Cash App. None is trust-minimized. Trust is math, not magic, and none of these three publishes the math. The stablecoin issuers behind the rails earn reserve interest on every dollar routed through, regardless of which app wins โ and the reserve backing underneath has never had a clean, independent audit that the whole industry quietly pretends isn't a problem. The apps, meanwhile, will grind each other's margins toward zero on a corridor Wise already commoditized, squeezed from both sides by crypto-native rivals and low-cost remittance incumbents.
The next 12โ24 months will be decided by licensing, not latency. Watch Krak's UK authorization status and its banking partners, not its fee page. Watch whether the on-chain savings ever reach the withdrawal button โ or vanish into a spread no table prints. The question was never whether crypto rails are cheaper. It is who collects the certainty when the savings disappear.