How we calculate your rewards

Our estimates lean conservative on purpose

When we can model a reward precisely, we do. When we can't, we lean toward under-counting rather than guessing high. That's a deliberate choice: an estimate you can verify against the issuer — and that doesn't over-promise — builds trust.

We're not claiming the number is a guaranteed floor. To keep estimates useful we do assume you'll realize some statement credits, and travel categories in particular can carry card-specific rules that shift your real earnings up or down. What we're promising is a bias toward caution, not a lower bound you're certain to beat.

How we estimate statement credits

Many cards come with statement credits — a monthly dining credit, an annual travel credit, a rideshare credit. Their headline value assumes you use every one, but almost nobody does. Counting them at full face value would overstate a card's worth and push you toward perks you'd never fully use, so we estimate them carefully:

• Not every credit is equally easy to use. We sort each card's credits by how easy they are to capture and assume you'll get more out of the easy ones than the fiddly ones: straightforward cash-like credits that simply lower your bill, coupon-style credits you have to remember to redeem on a schedule (a specific monthly dining or rideshare credit), single-merchant "lifestyle" credits that only have value if you already shop that specific brand (a gym, a clothing line, a specific rideshare membership), and soft perks like lounge access — which we count as $0 in the math even though they carry real value for some people.
• We learn from the credits you already use — but only to be more cautious, not less. If you tend to get less out of your current cards' credits than a typical person, we nudge a new card's credit estimate down. If you clearly get more out of them than typical, that just removes the extra caution — we still won't model a credit above the easy-case rate we'd expect for its kind, or above the credit's own face value. Before we've seen any of your usage we start from a deliberately conservative assumption rather than guessing high.
• You get one number, and you can change it. On a card we recommend we show a single "credits you'll likely use" figure, pre-filled from the model above. If you know you'll use more or less than we assumed, adjust it and the recommendation recalculates around your number.
• Once a card is in your wallet, you tell us which credits you'll use. We don't pre-fill coupon-style credits on a card you own — each stays off until you turn it on, so a card's credit value reflects only what you'll actually claim rather than a number we made up. The card shows how much it carries in credits (for example "up to $300/yr") to prompt you. Credits that post automatically with no effort — like an anniversary points bonus — are counted for you from the start.
• We recommend one card, and show how that changes with credit use. When we suggest a next card, the recommendation always uses the conservative, personalized credit estimate above — that single number is the basis for every headline and the dashboard figure. Alongside it we show two honest alternates: the best card if you ignore credits entirely (rewards alone), and the best card if you'd use every credit at full advertised face. The full-face view is an optimistic ceiling almost nobody reaches, so it's clearly labeled "not our recommendation" and never drives a headline. If a card looks worth it to you at a value we didn't assume, set your own number for it and re-evaluate — that value, not the advertised face, is what flows into your recommendation.

What our estimates may not include

To keep that promise, we leave a handful of hard-to-model perks out of the math rather than risk overstating them. Most of these mean your actual earnings could be higher than we show; a couple (travel fine print, how categories are defined) can go either way. Where they apply:

• Limited-time promotions. Short-lived elevated rates (e.g. a "5x on rides through September" deal) expire and vary too much to model reliably, so we credit the card's everyday rate instead.
• Fine print on travel rates and credits. We do model issuer travel-portal earn rates where a card offers them. What's harder to capture is the conditions attached: a travel credit that only counts when you book through the issuer's own portal (e.g. Capital One's $300 travel credit), or an elevated travel rate that only applies to airfare booked through the issuer's travel portal or directly with the airline — but not third-party sites like Expedia (e.g. the Amex Platinum 5x on flights). We generally credit the rate that applies to ordinary spend, so your earnings on these specific bookings may differ from our estimate.
• How categories are defined. We don't cover every possible spending category, and issuers don't agree on what falls into each one. Whether "entertainment," "communications," or even "travel" includes a given merchant — and whether transit or rideshare counts as travel — varies by card. We use reasonable mappings, but a specific purchase may be categorized differently by your issuer.
• Very-high-spend perks. Bonuses and statuses unlocked only after tens of thousands of dollars of annual spend on a single card touch very few people, so they're shown as notes rather than folded into your estimate.
• Single-retailer & store-brand bonuses. Extra points at one specific merchant (a chosen retailer, or an airline/hotel card earning on its own brand) don't map cleanly to a spending category, so we credit only the base rate on that spend.
• Rotating quarterly categories. Cards with 5% categories that rotate each quarter (and that you have to activate) appear in our catalog, but we don't factor their rotating bonus into rankings or recommendations. The eligible categories change every quarter, so crediting a fixed 5% would misstate what you'd actually earn — we credit the card's base rate instead.
• Certain redemption floors. A few cards have card-specific minimum redemption values that only matter for cash-preference holders of those exact cards; where the difference is tiny we leave it out.

We keep an internal registry of every one of these gaps and which direction each one pushes the estimate, so our omissions stay biased toward caution. Some of them — like rotating quarterly categories — we may simply leave out for the foreseeable future rather than model imperfectly.

How the math actually works

• You enter your spending — we never see your accounts. No bank linking, no card numbers, no aggregation. You tell us roughly what you spend where, and which cards you hold.
• We run every card in our current catalog against your profile. For each spending category we work out the best-earning card, at each card's real points value.
• "Best Next Card" is the single best add — not a full overhaul. Alongside our full suggested-wallet strategy, we also compute the one card that helps you most right now: once counting only reliable everyday earning, and once assuming you'll use a new card's statement credits too. It ignores the maximum number of cards you've told us you're willing to carry — we surface the opportunity and let you decide whether it's worth going over. You can also pick any card yourself and see exactly what it would win, with no such assumptions applied.
• An optional "cash-redeemable only" filter. If you'd rather not be shown new cards whose points are locked into a single airline or hotel program, you can restrict new-card suggestions to points you can take as cash or a statement credit worth at least a cent each. It only ever affects which new cards we suggest — cards you already own always keep earning.
• You decide what your points are worth. Choose a straight cash-back view or a maximize-value view, and optionally set your own per-program values on top. The estimate recalculates around whatever valuation you pick — including a single default rate if that's what you prefer.
• Catalog data is current as of August 2026. Card terms change often; we refresh the catalog regularly, but the issuer is always the final word.

The math is a starting point, not the whole decision

Rewards are as much about behavior and context as they are about math — and we can only model the math. We make the best attempt we can from the data available, but there's a lot we can't see and shouldn't pretend to.

A card might come out ahead on paper and still be the wrong call for you. Maybe the winner earns its edge inside an ecosystem you'd rather not join for the sake of a few thousand extra points a year. Maybe you'll happily accept slightly lower rewards for a card with better travel or purchase protection, a smoother app, or a lounge you actually use. Those are real, legitimate reasons — they're just not things a rewards calculator can weigh for you.

So treat our numbers as a well-researched starting point for your own judgment, not a verdict. The best card is the one that fits how you actually spend and live.

Fee and bonus dates are reminders, not exact deadlines

When you tell us when you opened a card, we estimate two timing signals: roughly when its next annual fee will post, and roughly how long you have to meet a sign-up-bonus spending requirement. Both are derived from your open date plus the card's typical terms — so an annual-fee date we estimate this way is shown with a "~", and a bonus window is our best guess at the deadline, not the issuer's official one.

Treat them as a nudge to go check, not the last word. Issuers set these dates precisely — often from your exact approval date — so they can differ from our estimate by days or weeks, and missing a real annual-fee or bonus deadline has real consequences. Always confirm the exact date in your card account.

Whether you can still earn a card's sign-up bonus

Separate from whether an issuer is likely to approve you, many cards limit their sign-up bonus — once per lifetime, once across a family of related cards, or only after a waiting period since your last bonus. When your logged card history suggests a card's bonus may be off the table, we flag it on the card so you're not counting on a bonus you can't get.

We try to be honest about how sure we are. Where the limit is provable from your own records or spelled out in the official terms, we say it more firmly; where it rests on widely-reported but unofficial behavior — like Amex's often-cited once-per-lifetime reset — we hedge it as a "you may not be offered this" heads-up rather than a hard no.

Two things to keep in mind. First, this is only as complete as the history you've entered — if you've held or canceled a card you haven't recorded, we can't account for it. Second, bonus eligibility is ultimately the issuer's call and can only be confirmed by applying. So treat the flag as a prompt to check, not a verdict either way. And note that we never let bonus eligibility change which cards we recommend or how we rank them — it's shown alongside a suggestion, never baked into it.

Your best-value wallet

Insights can show which subset of your current cards earns back the most after annual fees — spend rewards plus the statement credits you've told us you actually use, minus what each card costs to hold. We compare that "best-value" set against your full wallet today, and, if you've set a maximum number of cards you want to carry, against the best wallet limited to that many cards. This is exact, not a rule of thumb — we search every combination worth searching (pruning away only cards that are provably no worse to keep), so a smaller wallet never shows as more valuable than a larger one.

It's a review, never advice to cancel anything. It only counts credits you've confirmed you'll use, so a card can look worth dropping here purely because we don't yet know you use its perks — the tool links straight to confirming credits for that reason. It also can't see real-world keep-reasons like credit age, utilization, retention offers, or sign-up-bonus re-eligibility.

If you hold a Bilt card, its rent/housing value is included in this view — dropping it won't be recommended just because it looks ordinary on everyday spending alone. One residual approximation: the housing figure is measured once against your full wallet rather than re-measured for every possible smaller wallet, so in the "limited to N cards" comparison specifically, Bilt's true value could be a little higher than shown. That only ever means we understate Bilt, never overstate it, and it doesn't change whether Bilt itself is kept.

Always verify with the issuer

Card Strategist is an informational tool, not financial, tax, or legal advice. Sign-up bonuses, fees, credits, and earn rates change frequently and vary by applicant. Before you apply for a card or make a decision based on our estimates, confirm the current terms directly with the card issuer.

We've documented these exclusions to the best of our knowledge, but card rules and benefits change constantly and we won't have captured everything. Spotted something we've missed or misstated? The best way to reach us is the Send feedback button inside the app — we read every note, though we can't promise to chase down every edge case that affects only a handful of cardholders.