When the cheapest crypto card on paper is not the cheapest in practice

A worked example with numbers from the catalog.

Most crypto card listicles rank cards by headline cashback. I keep doing this exercise where I take two cards that look similar on paper, plug in real spending numbers, and watch which one actually ends up cheaper. The headline rate almost never wins.

Take a simple case. Card A advertises 4 percent cashback in the issuer loyalty token and 0 percent FX. Card B advertises 1 percent cashback in BTC and 1 percent FX. If you spend $1,000 a month, the ranked-list view says Card A wins by $30 a month.

Now read the fine print. Card A 4 percent applies only after staking $4,000 of the issuer token for 180 days. The cashback is paid in that same token, and the published 30-day average price has dropped 18 percent since the staking program launched. The token is also the only path to the 4 percent tier. There is no fiat fallback. The 0 percent FX has a footnote: foreign-currency transactions still pay a 0.9 percent crypto-to-fiat conversion fee at point of sale.

Card B 1 percent BTC cashback is uncapped. The 1 percent FX is what it says. There is no token exposure.

Now the same $1,000 a month, six months in. Card A holder paid roughly $720 in token-price drawdown on the staked position, earned 4 percent of $6,000 in cashback at average price, and lost 0.9 percent on whatever portion was foreign currency. Card B holder paid 1 percent FX on roughly half of spending and earned 1 percent BTC on all of it. The math, with reasonable assumptions, has Card B ahead by a wide margin by month four.

This is not an edge case. It is the modal case for cards advertising any cashback rate above 2 percent. I argued the general version of this in why most crypto card cashback rates mislead readers and have not changed my view. The follow-up question I keep getting is what to sort by then. I answered that one in a methodology note on which metric to compare 139 cards on, and the short version is total cost per dollar spent over six months, including conversion fees, FX, token drawdown if cashback is in a token, and any tier requirements expressed as opportunity cost.

The catalog where I track these numbers across 141 cards is at sweepbase.net/cards. I recently wrote up what I would ask any new crypto card founder pitching me a launch, which lists the five questions that almost always reveal whether the headline numbers will hold up. I also wrote up the operational lessons from running this catalog for three months for the dev.to crowd, which is a different angle on why I trust the underlying numbers.

There are three cards that survive this exercise consistently across a thousand-plus simulation runs at different spend levels. I am not going to name them here because the right answer changes depending on which currency you spend in, which region you can onboard from, and whether you actually want self-custody. The comparison calculator on the site lets you plug in your own monthly spend and see which cards beat which other cards at break-even.

If you remember nothing else from this post, remember the meta-rule: the higher the advertised cashback rate, the more likely it is paid in something other than the currency you spend in. That something is usually an issuer token whose price is correlated with the issuer marketing budget.