How we score every Canadian credit card. No bank influence.

Our principles

Four ideas drive every decision in our methodology. Disagree with them? Don't trust our rankings.

1. Math, not vibes.

The card with the higher TopRates Score has more total points after our rubric runs. Full stop. Our editors don't override the algorithm to "promote" a card we're paid more for. (If they did, we'd note it on the card page — and we'd lose the right to use the word "independent.")

2. Real dollars, not "up to" hypotheticals.

When we say "year-1 value of $728," we mean: at the StatsCan median Canadian household spend, what's the dollar return after the annual fee. No best-case scenarios where you have to spend $5,000 in the first month to unlock a welcome bonus that gets devalued in year 2.

3. The reader is the customer.

The bank pays us referral fees. The reader doesn't pay us anything. That's a structural problem we manage by: (a) running the algorithm blind to the referral payout, (b) publishing every input to the score, (c) flagging cards that aren't in our affiliate network so you know we have no financial incentive to recommend them.

4. We say what we wouldn't.

Most "best card" lists rank everything. We don't. There are 72 cards. We don't write reviews for the cards we'd never tell our family to apply for. We'll explain why on the methodology page (this one) — but we won't write a soft puff piece that ends with "Apply now ↓".

The 5-step process

How a card goes from "new product announcement" to "ranked on TopRates."

Weighting, in detail

We've talked to a lot of Canadians about what they actually care about in a credit card. The weights below reflect what consistently surfaces.

Rewards (30%) is the heaviest weight because, for most readers, it's the only thing that determines whether the card returns more dollars than it costs. We use a spending-profile-weighted reward velocity: a card that pays 5% on groceries is worth more to a grocery-heavy household than a card that pays 5% on transit, and the score reflects that.

Fees (20%) isn't just the annual fee — it's the foreign transaction fee, the supplementary card fee, the over-limit fee, and the cash advance fee, all blended. A card with a $0 annual fee but a 3% FX markup is not a "$0 fee" card to a traveller.

Insurance (15%) captures the full bundle: mobile device, rental car CDW, trip cancellation, out-of-province emergency medical, baggage, purchase protection, extended warranty. Dollar-coverage is part of the score; deductible isn't (because nobody knows their deductible until they file a claim).

Perks (15%) is the discretionary stuff: lounge access, Nexus credit, concierge, hotel status, free supplementary cards. We discount perks aggressively if they require triggering events most readers won't hit ("airport lounge access after $50k spend" — we score that as zero for typical readers).

Eligibility (10%) rewards cards with lower income requirements, faster approvals, and newcomer-friendly paths. It's a smaller weight because most readers self-filter to cards they'd qualify for.

Interest rates (10%) only matters if you carry a balance. We tag cards with "carries-a-balance friendly" if APR is under 12%, and the weight kicks up for those readers.

From points to dollars

The trickiest part of our scoring is normalizing points-based cards (Aeroplan, Scene+, BMO Rewards, Amex MR) against pure cash-back cards. Comparing 5% cash to "5× points" is meaningless without a points-to-dollar conversion.

We publish our point valuations and refresh them quarterly. Here's the table.

The 126 features, grouped

The full list is too long to display inline. Here are the 12 groups we track, with feature counts.

Data sources

Every input to our score comes from one of these sources. Issuer-published terms always win conflicts.

Editorial independence

We're an affiliate. The bank pays us a referral fee — typically $40–$180 per approved application — when a reader applies through our link. That payment does not flow into the algorithm. Here's how we structurally enforce that:

• The scoring engine runs in a separate code path from the affiliate management. The scoring code has no read access to the affiliate payout table.
• Editor compensation is salaried, not commissioned. An editor's pay doesn't change based on which card the reader applies for.
• Non-affiliate cards are ranked too. If a non-paying card wins its category, we publish that — and you'll see it labelled "No affiliate partnership — independent recommendation" next to the score.
• The data refresh is automated. When an issuer updates terms, our scrapers pick it up and rescore. An editor can't quietly hold a stale score because the affiliate payout went up.

The team

Three people own the credit-card vertical. Two are credentialed; one is an editor.

Algorithm changelog

Every change to the scoring methodology is logged here.

See the full changelog →