The annual-fee break-even method

The renewal notice arrived and you did what you always do: you thought about the one big credit you used last year, felt vaguely annoyed, and paid the fee. Or you canceled in a huff and spent six months wondering whether that was dumb. Both moves were vibes. Neither was math.

The method is this. Price everything the card gives you at what it is actually worth to you: hard credits you will genuinely use, plus the extra earn above a free card, plus perks you would genuinely pay for. Add them up, compare against the fee, and demand a cushion, because one of those credits will get skipped next year the way they always do. Everything competes against a no-fee 2% cash-back card, because that card exists and it is free. When the total comfortably clears the fee, keep the card. When it does not, you have a downgrade or a phone call to make, and the size of the gap tells you which.

We will run the whole method on one card, a $395-fee travel card, start to finish. The arithmetic transfers to an $895 card, a $95 grocery card, or a $650 airline card unchanged.

The method in one paragraph

Hard credits at face value, but only the ones you would buy anyway. Earn delta: what the card pays you on your real twelve months of spending, minus what a free 2% card would pay on the same spending. Perks at the price you would have paid out of pocket, zero if you would have walked past. Total those three, divide by the fee, and read the ratio: below 1.0 means you are paying for the privilege, and 1.2 to 1.5 is a sensible comfort band, not a law of nature. That is the entire machine. The rest is discipline about each number.

1Count hard credits at the price you would actually pay

A hard credit is statement money against a purchase. The rule is brutally simple: if you were going to spend that money anyway, the credit counts at face value. If the credit requires you to buy something you would not otherwise buy, it counts at zero. Not at a discount. Zero.

Credits also differ in friction, and friction predicts whether you will actually collect. Rank every credit on the card into one of four tiers.

Real examples, from current benefit structures. A $300 annual travel credit that applies to travel you already book, like the one on our example card through the issuer's travel portal, sits between automatic and easy: if your travel routing tolerates the portal, count it at $300 (Capital One product page, current as of August 2026). Chase's Sapphire Reserve travel credit, $300 on an anniversary-year basis, behaves much the same for anyone who buys flights. Contrast the enrollment-gated kind: American Express Platinum's $400 Resy credit arrives $100 per quarter only after enrollment, and its $300 Digital Entertainment credit pays $25 monthly against enrolled subscriptions (American Express benefit pages, August 2026). Those can absolutely be worth face value for a household that already orders those things, but "already" is the load-bearing word.

Then there is the prepaid-gated tier, the one that fools people most. The Platinum's $600 hotel credit requires prepaid stays in specific collections with a two-night minimum at one property type. If that matches how you travel, it is real money; our full credits audit walks that ledger line by line. If it does not, the marketing total, "over $3,500 in value," is a number for somebody else's life.

Back to our example card, step one output: the $300 travel credit counts at $300, because this household books about $1,400 in travel a year anyway. The $120 Global Entry or TSA PreCheck credit counts at zero this year, because nobody in the house is renewing an application; in a renewal year it is worth $120 once, not annually, and amortizing a five-year credential at $24 a year is honest but tiny. The card's 10,000 anniversary miles count at the issuer's own floor, roughly $100 toward travel booked through the portal, because that floor needs no optimism to realize.

Step one total: $400.

2Measure the earn delta

The card does not exist in a vacuum. Its earn rate only matters relative to what a free card would pay on identical spending. The formula: card earnings on your actual twelve-month spend, minus free-card earnings on the same spend. Only the difference belongs to the fee card.

Here is the sample household we will carry through the rest of the article. Label it clearly: this is a scenario, not a recommendation, and the numbers are chosen to be ordinary.

The earn structure, 5x on portal bookings and 2x everywhere, with miles worth a cent toward portal travel, comes straight off the card's product page (Capital One, August 2026). The 2% baseline is the standard flat-rate no-fee card; Wells Fargo's Active Cash is the archetype, stable at 2% for years.

The delta is $90 a year. Notice how small that is. A 2x-everywhere card earns exactly what a 2% cash card earns once you value miles at the issuer floor, so the entire earn advantage lives in the portal bonus category. This is the pattern on most fee cards: one or two categories carry the delta, and the rest of your spending is a tie. Households that never book through the portal have an earn delta of roughly zero here, and the fee decision falls entirely on credits and perks.

3Price the perks at what you would have paid

Perks are not credits. Lounge access, hotel status, fee waivers, free-night certificates: none of them post as money, so each gets one honest question. What would I have paid for this, out of pocket, this year?

Lounges first, because they carry fee cards emotionally. If you take four visits a year and would genuinely have bought a quiet seat, a snack, and Wi-Fi for $30 a visit, the perk is worth $120 to you. If you would have walked past and eaten at the terminal deli, it is worth $0 regardless of what a visit "retails" for. Six visits at your honest price beats ten visits at a fantasy price every time. The lounge rules changed across every network in early 2026, guest fees especially, so re-price this line before any renewal; our lounge reset guide has the current numbers, and the post-change renewal math for this exact example card runs separately.

Status counts only when you would have used it: a mid-tier hotel status you never notice is decoration. Free-night awards count at the cash rate of a night you would actually have booked, capped by the award's point ceiling; a certificate that expires unused, which happens constantly, is worth its face only in the year you redeem it. The Global Entry application credit already appeared in step one as zero, so it does not get a second life here.

For our example household: four lounge visits at a self-priced $30, so $120. Status on this card is thin, call it $0. Step three total: $120.

4Value the points at three different prices

Every points currency has three values, and sloppy articles mix them constantly. The issuer cash-out floor is what the issuer itself will give you, no creativity required: our example card's miles at a cent each toward portal travel, Chase Ultimate Rewards at a cent each in cash. Community estimates run higher when you transfer points to airline and hotel partners; for Membership Rewards, community commentary through early 2026 still describes 1.5 cents or more per point as achievable on the right transfers (travelupdate.com, March 19, 2026, a community estimate, not a promise). Your realized value is what your last three actual redemptions paid you, and it is the only one of the three you can verify.

The discipline: run the break-even at the floor. If the card clears its fee at the issuer floor, the answer is robust, because partner redemptions only add upside. If the card only clears at a community estimate, the card is a bet on your own redemption behavior, and the bet needs a plan attached: specific trips, specific programs, dates you would actually fly. Never let a dream business-class redemption do the work a boring floor value should do.

One more points rule belongs in this method even though it belongs to a different decision: points are the reason downgrading usually beats canceling. Closing the only card in a currency can forfeit the balance outright, which turns a marginal fee decision into a permanent loss. The mechanics live in our points-on-cancellation piece, and they matter more than any line item above.

5Compare against the fee, with a margin of safety

Assemble the example card's year.

A 1.55x ratio with the lounge line priced honestly is a comfortable keep. Now stress it, because a single year of changed behavior should not flip the answer. Zero lounge visits, which happens to frequent travelers in a light year: $490, a 1.24x ratio, thinner but survivable. The travel credit goes unused because the household books a package trip through a different channel: $290, a 0.73x ratio, and the card has crossed from keep territory into downgrade territory. Re-run the spend table at plus and minus 20% and the earn delta moves between roughly $72 and $108, barely shifting the outcome, because credits, not earn, dominate this card's value. That is the diagnostic: when credits dominate, the stress test that matters is behavior, not spending. When a card's value is mostly earn delta, the spend swing is the one to model.

The margin band, 1.2 to 1.5, is prudent practice rather than regulation. It exists because life happens: a credit gets forgotten, a lounge gets skipped, a hotel program devalues. A card sitting at 1.05x is one missed credit away from being a donation to the issuer.

6Run everything against the free baseline

The last step is the one people skip, and skipping it flatters every fee card on earth. Somewhere in your wallet sits a card that charges nothing and pays 2% on every purchase. It tracks no credits, respects no portals, and asks no renewal-day questions. On the scenario spending above it quietly earns $480 a year.

So the question is never "is this card worth $395?" It is "does this card beat a free card by more than $395?" The example card, valued entirely at the issuer floor, pays about $570 in earn plus $520 in credits and honestly priced perks: $1,090 all in. Its advantage over the free card is $610, which is exactly the number the table produced, because step two already measured earn as a delta rather than a gross. The fee consumes $395 of that advantage, leaving roughly $215 a year. That thin slice is what the whole annual relationship actually buys. For some households the lounge access or the portal bonus rates clear the bar comfortably. For many, this paragraph is the one that finally talks them into the downgrade path.

Where the method usually fails

Three failures account for most broken break-evens. Counting credits at face value that expire unused; the fix is auditing your last twelve statements for what actually posted. Valuing points at the best redemption you ever had; the fix is the floor-plus-plan rule from step four. And ignoring the free baseline; the fix is step six. A fourth failure is subtler: forgetting that the fee itself can be negotiated. If the ratio lands in the gray zone, the retention desk can close the gap, and your break-even number is your asking price. A 25,000-point offer is worth $250 at the issuer floor and about $375 if you reliably hit 1.5 cents on transfers; the range is the point, so decide before you dial which end you believe. The timing and script for that call have their own guide.

What the number tells you to do

A ratio at or above your comfort band: keep the card, re-run the math every year, and stop feeling weird about the fee. Below 1.0, or in the gray zone with no retention rescue: the decision becomes which exit, and that is a framework with three doors, points safety and credit age built in. The arithmetic you just did is the input it needs.

Run your own card through our annual-fee calculator with last year's real statements, and the keep, downgrade, or cancel answer falls out of your numbers instead of your mood.