In 2026 Mastercard raised its fee for excessive authorization attempts from $0.10 to $0.50 per retry, a five-fold increase meant to slow down card-testing fraud. Visa moved in the same direction. As of January 1, 2026, acquirers have to keep their VAMP ratio below 0.5%, where that ratio counts fraud and disputes against card-not-present volume. Both changes raise the cost of getting authorization wrong, and interviewers for fintech PM roles now ask about authorization rate.
If you are preparing for a payments PM interview, this guide covers what authorization rate is, why payments get declined, and how a PM moves the number without inviting more fraud.
What authorization rate measures
Authorization rate is the share of payment attempts that win issuer approval. If a merchant sends 1,000 card charges and issuers approve 940, the authorization rate is 94 percent. The gap is made up of declines, and each decline is either a sale the merchant loses or a customer who tries again at another merchant.
The number sounds simple, and the detail is the focus of most interviews. A useful answer separates hard declines from soft declines. A hard decline means the issuer will not approve the charge no matter how many times you retry, for example a closed account or a stolen-card flag. A soft decline is temporary, such as insufficient funds or a network timeout, and a later retry can earn approval. PMs treat these two groups differently because each group needs a different fix.
Why payments get declined
Declines come from several places, and naming them shows you understand the flow. The issuer is the most common source. Issuer risk models decline a charge when something looks off: a new device, an unusual amount, a billing country that does not match the card, or a sudden burst of attempts. Some of these are real fraud, and some are real customers caught by a cautious model. Those second cases are false declines, and they cost merchants real revenue.
Other declines come from stale card data. Cards expire, get reissued after a breach, or change numbers, which leaves the saved card as a dead record. A subscription business feels this every month as involuntary churn. Still other declines come from the merchant's own setup, such as a wrong currency, a missing field, or a retry pattern that trips a fraud filter.
How a PM raises the number
Once the decline reasons are clear, the levers come into view. Network tokenization replaces the stored card number with a token tied to one merchant and one card, and the token updates itself at the next card reissue. Visa reports about a 4.6 percent authorization uplift on card-not-present transactions that use network tokens, and Mastercard reports about 2.1 percent. An account updater service does a related job, refreshing stored card numbers after each change. Many teams run both tools.
Smart retries are another lever. Instead of retrying a soft decline right away, the system waits and retries at a time the issuer is more likely to approve, such as after a payday for an insufficient-funds decline. Sending richer data with each request also helps, because issuer models approve more often when they can see a cardholder name, an address, and a prior token. Processors package these tools under names like Stripe Authorization Boost and Adyen Uplift, and a PM should know what sits inside the box rather than the brand name on the box.
The fraud trade-off
Here is the part that belongs in a full answer. You can lift authorization rate inside a dashboard by approving riskier charges, and the fraud and chargeback cost lands the next month. The new Visa and Mastercard rules put a number on that cost. If approving more charges pushes a merchant's VAMP ratio above 0.5%, the acquirer faces fines on fraud and dispute transactions, and aggressive retries on dead cards now cost more per attempt under Mastercard's higher fee.
A payments PM does not chase authorization rate in isolation. The metric travels with a fraud rate, a chargeback rate, and a cost-per-attempt figure. A complete answer in an interview names that tension out loud and explains how you would hold fraud flat while authorization climbs, usually by going after false declines and stale data first, since those add revenue without adding risk.
How this shows up in interviews
Expect three shapes of question. A metric question asks how you would diagnose a drop in authorization rate from 92 percent to 88 percent over a week. Walk through segmenting by issuer, card type, decline code, and geography until the cause narrows to one bucket, the same way you would debug any metric drop. A design question asks you to build a billing system for a subscription product and to keep payments flowing past card expiry. Bring up network tokens, account updater, and retry timing. A trade-off question asks how you balance approval and fraud, and that is where the VAMP and retry-fee rules earn their place in your answer.
Interviewers at processors and platforms care less about memorized statistics and more about whether you reason from the cardholder and issuer point of view. A decline is a moment where a real person wanted to pay and hit a wall. Frame answers from those two sides, and the rest is detail.
How to prepare
Learn the decline taxonomy first, because it anchors every other answer. Read the public docs from Stripe, Adyen, and the card networks on authorization optimization, since they spell out the levers in plain terms and use the same vocabulary as your interviewer. Keep one number range in mind for tokenization uplift so you can speak in specifics, and know the 2026 rule changes well enough to explain why the fraud side of the trade-off got more expensive this year.
Then practice one end-to-end story. Pick a checkout or subscription flow, state a starting authorization rate, name the biggest decline bucket, choose a lever, and say how you would measure whether it worked while watching fraud. That single worked example will carry you through most of a payments PM loop, and it shows the thing interviewers want to see: a PM who can move a hard metric without breaking the surrounding system.