Credits you don't use aren't credits. Count only the ones you'd have bought anyway.
That's the first rule. The fee is printed right there in the offer. The break-even isn't printed anywhere... because it depends on a number only you have. What you spend, and where.
I did the division. Here's what it looks like.
What The Offer Puts On The Front Page
The CFPB cites one study of sampled card offers. Rewards showed up on the front page over 94 percent of the time. Annual fees: 78 percent. APRs: 28. Late fees: 6 (CFPB, May 2024).
Read that again. The thing you earn is the loudest line. The thing you pay is quieter.
And the terms aren't locked. The same CFPB report says many of the largest issuers "reserved the right to change their rewards program at any time" (CFPB, May 2024). In many cases without notice. It also says issuers are increasingly limiting access to experience benefits like airport lounges.
Then there's the fee itself. In December 2025 the CFPB reported the average annual fee on general-purpose cards rose 21 percent, from $105 in 2022 to $127 in 2024, and said premium products drove it. 16 percent of accounts charged an annual fee in 2024. That's an average across general-purpose cards. It's not what any specific card charges, so don't use it as one.
Either way... the rewards pitch is loud and the math is yours to run.
The Formula, In Plain Words
Premium net is the rewards you earn, plus the credits you'll really use, minus the fee. Flat-card net is your spend times the flat rate. Break-even spend is the fee minus the credits, divided by the premium card's blended rate minus the flat rate.
Three lines. No app required. But the inputs are the whole fight, so run your own.
Enter the fee, the credits you'd really use, and what you spend in each category. The rate on each row is your number, not the issuer's. Perks like lounge access and insurance aren't in the math. You price those.
Tool's also at /calculators/finance/card-fee-breakeven if you want it full-page.
Fixture A: The Flat Card Wins By $300
Full inputs, because a number without its inputs is a rumor. Fee: $500. Credits: $200 face value, used 50%, so $100 counted. Flat-rate card: 2%. Spend: $20,000 at 3% plus $10,000 at 1%.
Rewards come to $700. Add the $100 of credits, subtract the $500 fee, and premium nets $300. The flat card on the same $30,000 nets $600. Flat card ahead by $300.
Blended premium rate: 2.33%. Barely above the flat rate, so the gap closes slowly. Break-even is $120,000 of spend. You'd need four times the spend you entered before the premium card catches up.
Is that a bad card? I have no idea. That's not what the math says. It says this fee, with this credit haircut, on this spend, loses to a plain 2% card by $300.
Fixture B: Same Fee, Different Spend
Same fee. Same $200 of credits used 50%. Same 2% flat card. Only the spend moves: $50,000 at 4%.
Rewards: $2,000. Premium nets $1,600. Flat card nets $1,000. Premium ahead by $600. Break-even: $20,000.
Nothing about the card changed between A and B. Spend concentrated where the rate is high turned a $300 loss into a $600 win. That's the entire lesson, and it's why "is this card worth it" is the wrong question. Worth it for which spend?
The Verdict Flips (Even Without A Break-Even Number)
Here's the case that trips up bad copy. Fee $100. Credits $300, used 100%. Flat card 2%. Spend $10,000 at 1%.
Premium nets $300. Flat card nets $200. Premium wins by $100.
The premium rate is below the flat rate, so the calculator prints no break-even number. Doesn't matter. The verdict is still there. Run the spend up and it flips: level at $20,000, both cards net $400. At $30,000 the flat card wins by $100 ($500 vs $600).
So: premium ahead below $20,000, level at $20,000, flat ahead above it. Credits cover the fee, then the lower rate eats the lead as spend grows.
I'm not going to hand you a blanket verdict on premium cards. Blanket verdicts are lazy. The verdict branches on your spend, and that's why you run it.
When The Paid Option Is Actually Worth It
Not a pitch against paying a fee. Cases where the math holds:
- Credits you already buy. If the credits cover spending that was already in your budget and you'll use them in full, they're near cash. That's the $100-fee case above.
- Spend concentrated where the premium rate is high. Fixture B breaks even at $20,000 of spend. Fixture A's 2.33% blended rate needs $120,000.
- Perks at your valuation. Lounge access, insurance, status... none of that is in the calculator's number. If you use it, put your own price on it and add it by hand. Nobody else's number counts here.
- You pay in full. This one's not optional. The CFPB reports that cardholders who revolve debt pay 94 percent of total interest and fees but gain less than 30 percent of rewards benefits (CFPB, May 2024). If you carry a balance, you need an interest calculator before you need this one.
- The terms hold. A break-even computed today assumes today's program. See "reserved the right to change their rewards program at any time" above.
One more thing, so the baseline isn't hypothetical. As of 2026-10-11 a no-annual-fee card exists that pays 1% when you buy and 1% as you pay, in points, if you make at least the minimum payment on time. No bonus, no APR math, no endorsement. It's an existence proof that the 2% flat-rate baseline in Fixtures A and B is something a person can actually hold.
The CFPB counted more than $40 billion in rewards earned in 2022 on major issuers' general-purpose cards (CFPB, May 2024). It also says issuers may cover that cost through other revenue channels, such as annual fees.
If the flat card wins on your numbers, you just saved a fee. If premium wins, you know the exact spend you have to keep up. Same exercise as days to FI, same instinct as the calculator that calls you out... turn the marketing line into a number you can check. It's all a math problem.
The fee is printed. Go print the other number.