Crypto card fees: map the full cost before you compare rewards

Funding can include an on-chain network cost, a provider deposit rule, a card purchase cost, or a bank-transfer condition. Spending can involve a balance conversion, a card programme charge, an FX charge, and an exchange-rate spread. They should not be presented as one generic 'card fee'.

Write down the funding asset and network, then the merchant currency. That simple distinction prevents a low-cost deposit method from being mistaken for a low-cost purchase route.

Provider fees and network fees should occupy different rows. A provider may charge zero for receiving a supported stablecoin while the originating wallet still pays network gas. The same provider may charge for withdrawing the balance later. Both affect total ownership cost even though neither is a merchant card fee.

Conversion can be explicit or embedded in a quote. Record the reference price or rate shown before confirmation and the amount that actually reaches the spendable balance. Calling a conversion 'free' does not establish that the quote has no spread, just as a visible percentage does not describe every part of a rate.

A percentage can be limited to non-USD spend, a defined card tier, a particular country, a reward season, or a capped amount of eligible purchases. The condition is part of the price. If the page does not say when a number applies, locate the detailed fee guide or treat the estimate as incomplete.

For example, a provider may publish a no-FX position for a specified set-up while another provider documents a non-USD FX range. Neither statement tells you the total result until you match it to your funding and spending path.

Base currency matters. KAST documents 0% card spend in USD and a 0.5%–1.75% fee for non-USD transactions. Gnosis Pay says it adds no FX fee and applies Visa's exchange rate. Ether.fi currently labels its zero-FX treatment for EUR purchases as beta. Those statements describe different scopes and cannot share one 'FX fee' cell without qualification.

Dynamic Currency Conversion is a separate merchant or ATM offer. Visa explains that DCC can include its own exchange rate and markup and must be presented as a choice. Selecting the card's home currency can therefore add a cost even when the issuer itself publishes no added FX fee.

Rewards should not cancel a fee automatically. Ask what asset is paid, whether it can be used immediately, whether a paid tier or held token is required, what purchases qualify, and whether a cap applies. The headline rate is a scenario input, not a conclusion.

A transparent comparison shows the cost first and then explains the assumptions required for a reward to offset it. That protects users from choosing an expensive route to chase a maximum that does not match their usual spend.

Normalize the reward into a conservative spendable value. Cash credited to the same balance, points with a provider-defined value, and a volatile token are not equivalent. Apply caps and excluded merchant categories before multiplying a headline rate by monthly spend.

Dated programmes require a stop date in the model. Gnosis Pay's current interim cashback programme runs through September 30, 2026 and ties rates and weekly caps to GNO holdings. A comparison that carries that rate forward indefinitely would turn a disclosed promotion into a false permanent feature.

For one realistic month, list deposits, conversion events, card purchases, foreign-currency purchases, cash withdrawals, and tier charges. Add the relevant documented cost beside each event. Keep uncertain values marked as unknown rather than filling them with an optimistic estimate.

Repeat the ledger when your travel pattern, spending currency, or chosen tier changes. The result is a personal cost comparison that is more useful than a generic table of fee labels.

Use three columns for every line: documented amount, scope, and source date. A 1% fee with no identified base is not ready for arithmetic. Neither is an 'up to' reward without its cap, qualifying balance, eligible spend, and payout asset.

Run a low, expected, and high-cost case when a fee is expressed as a range or a spread cannot be known in advance. The goal is not a false single number; it is to see whether one uncertain input can reverse the decision.