A card with a fee is not more expensive than a free card. It is more expensive until a spending threshold, and cheaper after it. The only question worth asking is where that line sits and whether you are on the far side of it.
Of the 98 personal cards we track, 44 charge no annual fee, 19 charge under $100, 19 charge between $100 and $149, and 16 charge $150 or more. The most expensive is $799. The middle of the range is $35.
The Arithmetic Is One Line
A fee card is worth paying when the extra it earns over the best free card exceeds the fee. Not when it earns more than nothing, which is the comparison the advertisement invites. The break-even is:
Annual fee, divided by the difference in earn rate, equals the spending needed to break even.
The difference is the part people get wrong. Across the 32 cash back cards we track, the best base rate on a free card is 1.5% and the best on a fee card is 2%. The difference is 0.5 percentage points, not 2. So a $120 fee needs $24,000 of annual spending on that card to break even, which is $2,000 a month, every month, on one card.
| Annual fee | If the card earns 0.5% more | If it earns 1% more | If it earns 2% more |
|---|---|---|---|
| $39 | $7,800 a year | $3,900 a year | $1,950 a year |
| $99 | $19,800 a year | $9,900 a year | $4,950 a year |
| $120 | $24,000 a year | $12,000 a year | $6,000 a year |
| $150 | $30,000 a year | $15,000 a year | $7,500 a year |
| $799 | $159,800 a year | $79,900 a year | $39,950 a year |
Read the top right of that table and the bottom left together. A modest fee against a genuinely better rate clears easily. A large fee against a half point edge does not clear on any realistic household budget, and the cards in that corner have to be justified by something other than the earn rate.
Four Ways the Arithmetic Gets Flattered
The Bonus Category Rate Is Not the Card Rate
A card advertising 5% is paying 5% on one category. Across the cards we track, the middle base rate is 1 point or percent per dollar, while the middle grocery rate is 2. The rate that decides what the card is worth is the blended rate across everything you actually buy, and it is always closer to the base rate than to the headline.
The Cap Is Usually Not Mentioned
Only 42 of the 98 cards publish an annual cap on their bonus earning, and where one is published the middle figure is $25,000 of spending. A card paying 5% on groceries up to $6,000 a year pays 5% on the first $6,000 and its base rate on everything after. If you spend $12,000 on groceries, the effective grocery rate is not 5%. Where a cap exists and is not published, the break-even you calculated is optimistic by an unknown amount.
The Welcome Bonus Only Happens Once
Sixty eight of these cards pay a welcome bonus, and the middle one requires $2,000 of spending within three months. Seventeen cards also waive the first year fee. Both are real value and both are first year value.
A card that is worth $500 in year one and loses $60 a year after it is a card that turns negative in year nine, and every “first year value” figure published anywhere includes the bonus. The honest test is the second year: strip the bonus and the waived fee out, and see whether the card still clears its own fee on your ordinary spending.
Points Are Worth What You Assume They Are Worth
A break-even in points depends entirely on the valuation applied to them, and valuations are assumptions rather than facts. We value Membership Rewards and Aeroplan at 2.0 cents and issuer fixed programmes at whatever the issuer pays, which across our data runs from 0.1 cents to 2 cents a point. At 2 cents a card can look like it pays 10% on groceries. At the cash-out rate the same card pays a fraction of that.
If the points can only be spent one way, or only with one retailer, or only through one travel portal, the valuation should reflect that rather than the best case redemption somebody achieved once. The comparison has a switch that re-ranks every card at cash-out value in one click, and how we value points sets out the assumption and three ways it can be wrong.
When a Fee Is Worth Paying Regardless of the Rate
The break-even calculation only prices the earn rate, and some fees buy something the earn rate cannot express. These are worth paying for if you use them and worth nothing if you do not:
- No foreign transaction fee. Of the 73 personal cards that publish a foreign transaction fee, 59 charge 2.5%. Only 5 publish 0%. Somebody spending $8,000 a year in another currency saves $200 on the fee alone, which clears most annual fees before a single point is earned.
- Travel medical insurance. Priced separately it is not cheap, and the coverage terms and age limits on a card vary enough that the certificate rather than the marketing page is the thing to read.
- Free supplementary cards. 52 of the 88 cards that publish a figure charge nothing for an additional card, and the most expensive charges $250. For a household putting two people on one card, that is a real line.
- Lounge access, but only sometimes. On several cards the membership includes zero free visits and each entry is charged. A membership with no visits is not lounge access, and this is one of the most consistently overstated benefits in the category.
How to Check Your Own Number
- Take your real annual spending, by category, from a year of statements rather than from an estimate.
- Work out what the best free card would pay on it.
- Work out what the fee card would pay on it, applying the caps.
- Subtract the fee. If the answer is not comfortably positive in an ordinary year with no welcome bonus, the free card wins.
The comparison does exactly this once you enter your own spending, across all 98 personal cards, and it ranks on the result rather than on the headline rate. If you would rather start from the answer, the cards with no annual fee are listed separately, and how to read an earn rate covers the caps in more detail.
The Benchmark Weekly
One email a week. Elevated welcome offers while they are still open, new cards and accounts as they arrive, and any fee, rate or earn rate that moved, with the old figure and the new one. A quiet week gets no email.