How CardWise scores a credit card

1. What goes in

Your monthly spending by category, the preferences you set, and — only if you choose — your age band, employment and approximate income. Nothing is sent to a server, stored, or put in the address bar. The plain-language box is read by rules in your browser; it is not a language model, and you can correct what it understood before anything is calculated.

2. Hard requirements come first

A card is removed, with the reason shown, when it is not open to new applicants; when its official terms are too incomplete to value; when it charges a fee and you asked for lifetime free only; when its fee is above your limit; when you need RuPay and it has none; when you want cashback only and it pays points; or when a published eligibility criterion appears not to be met. When a removed card would otherwise have ranked in the top three, the results say so.

Some cards are issued only to members of one body or force: ICAI’s card for chartered accountants, the cards for ICSI and ICMAI members, and the cards for the Army, the Navy, the Coast Guard and the Assam Rifles. Each such card quotes the issuer’s sentence saying who may apply. It is ranked only when you say you belong, in the eligibility step or in your description (“I am a chartered accountant”); otherwise it stays out of the ranking and out of the “would have ranked” note, but the search still finds it.

3. Net annual value

Net value (from the second year) = rewards after caps and exclusions + cashback milestones + fuel surcharge waived − annual fee after any waiver, with GST − forex markup with GST − redemption fees. The first-year figure uses the joining fee instead of the annual fee and adds welcome benefits that have a stated rupee value. Vouchers earned at milestones are listed apart, as potential value, because they depend on whether you use them. Lounge access, insurance and other perks are not given a rupee value.

• Money is calculated in whole paise and rates in basis points, so there is no floating-point rounding in any figure.
• Each category of spend is matched to the most specific reward rule the issuer states — Amazon before “online”, “online” before “all other spends”. Spends in categories the issuer excludes earn nothing. A category with no verified rule earns nothing and is listed as missing, not guessed.
• Caps are applied per their period (statement cycle, quarter or year) and across all the rules that share them. Rewards above a cap are not paid unless the issuer says the excess earns the base rate.
• Points are valued at the issuer’s stated rupee value per point. One redemption a year is assumed where a redemption fee applies.
• The fee waiver counts your yearly spend on the card, less the categories the issuer says do not count; where it does not say, every spend is counted and the assumption is shown.
• GST on fees is the rate the issuer states, or the standard 18% where it says only “applicable taxes”.
• Monthly figures are multiplied by twelve. Minimum transaction sizes for a rate or for the fuel waiver are assumed to be met, and the assumption is shown.

The comparison assumes the full statement balance is paid by the due date. Interest on a revolving balance and late payment charges are not netted off; they usually exceed any reward.

4. Personal Fit Score

A score out of 100, calculated in code from five parts with these starting weights: net value 40, category match 25, fees and restrictions 15, benefits you value 10, preference fit 10. A part that does not apply to you — no benefits asked for, no preferences set — is left out and its weight shared among the others. The score is shown rounded to the nearest 5, because the evidence does not support finer distinctions.

• Net value: net value as a share of your yearly spend; 4% or more scores in full, a loss scores zero.
• Category match: for each category, the reward rate the card actually pays after caps, against a full mark at 5%, weighted by your spend there.
• Fees and restrictions: half for the annual fee after waiver (zero at ₹5,000 or more with GST), three tenths for the share of your spend the card excludes or cannot take, one fifth for the share of rewards lost to caps.
• Benefits you value: the share of the benefits you asked for that the card provides on its official terms; a spend-gated lounge benefit counts half.
• Preference fit: the share of your stated preferences the card meets.

Missing information never scores in full: a category without a verified rule earns nothing, and an unknown forex markup cannot meet a low-forex preference.

5. Evidence Confidence, separately

From the quality of the sources (official issuer documents only), how long ago they were checked (fresh up to 120 days, ageing to 240, then stale), whether every term relevant to your spending is on record, whether points must be valued, how many assumptions were needed, and any disagreement between issuer documents. High confidence: fresh, complete for your spending, no assumptions beyond the standard ones. Moderate: one gap, ageing, points, or several assumptions. Limited evidence: two or more gaps. Needs verification: stale, disputed, or not open to new applicants.

6. Eligibility, separately

Only criteria the issuer publishes — age, income, employment, a fixed deposit — compared with what you entered. The outcome is one of: published minimum criteria appear to be met; one or more appear not to be met; insufficient information to assess; or the issuer does not disclose its requirements. No probability of approval is given, because no validated data supports one.

7. How the card data is kept

Each card record cites an evidence file holding the source documents — official URL, title, type, date, when checked and a fingerprint of the text read — and, for every figure, the exact sentence it was copied from. The build checks that each figure appears in its quote and that each source is on the issuer’s own domain, and refuses to publish a record that fails. Superseded catalogues are kept byte-for-byte with checksums, and a changed value without a change-log entry fails the build. The research dashboard lists stale records, gaps and the change log.

8. The search box

The box above the questions takes anything: a card or bank name (one typo is forgiven, two in a long word), a benefit such as lounge access, lifetime free, low forex markup, movies or golf, a kind of spending, a fee limit, an income, or amounts. Words that name a benefit or a limit filter the cards; amounts on a kind of spending rank them by the net value worked out above; other words are matched against card and bank names. Each reason shown beside a card is text from that card’s record, quoted from the issuer, never an echo of what you typed. If no card has everything asked for, the closest are shown with what each lacks. It runs in your browser; nothing typed is sent anywhere.

9. What “popular” means

Two official signals only. Issuers are ordered by the number of credit cards each has in use, from the Reserve Bank of India’s monthly bank-wise card statistics (ATM/POS/Card Statistics - August 2026). A card is marked popular only when its issuer labels it so (“Best Seller”, “Popular”, “Trending” and similar) on its own website; the label is quoted with its source. No popularity is inferred from search trends, social media, sales claims by others or this site’s own traffic. Popular cards

10. What it will not do

• Invent a fee, rate, cap, eligibility rule or source.
• Rank by commission or payment: there are no affiliate links.
• Promise approval or give an approval percentage.
• Ask for PAN, card numbers, CVV, OTPs or bank passwords, or run a credit check.
• Recommend borrowing, carrying a balance or applying for credit you do not need.