Nilson Report

Issue 1181 | Aug 2020


Companies featured in this issue include:

Top 7 Acquirers in Latin America—2019 vs. 2018

10 Largest General Purchase Acquirers in Latin America

45 Largest Visa/Mastercard Merchant Acquirers in Latin America

U.S. General Purpose Cards—January 1-June 30, 2020

Market Shares of Purchase Transactions

U.S. Credit and Debit Cards at Midyear 2020 vs. 2019

U.S. Store Credit Cards—2019 with Change vs. 2018

Investments and Acquisitions—July 2020

U.S. General Purpose Cards

The table on page seven shows Visa, Mastercard, American Express and Discover card credit and debit card purchase volume (spending) for cards issued in the U.S. for the first six months of 2020. Figures for credit card purchase volume are shown here.

1. Visa, Credit
$909.38 billion purchase volume, -9.2%
2. Mastercard, Credit
$395.58 billion purchase volume, -8.2%
3. American Express, Credit
$329.55 billion purchase volume, -18.0%
4. Discover, Credit
$64.71 billion purchase volume, -7.0%

Full access to the U.S. General Purpose Cards results is available when you subscribe to the Nilson Report.



Visa Adds Artificial Intelligence to Stand-In Processing

Every card issuer in the world using VisaNet to process card transactions receives backup from the network when their own transaction authorization systems are unavailable owing to normal maintenance or a problem such as an outage. 

Stand-in processing (STIP), which is as old as VisaNet, occurs in milliseconds. It exists to protect the reputation of the issuer and the card brand by assuring that cardholders receive a response after having swiped, inserted, tapped, or inputted a card. 

Issuers create their own parameters for STIP based on factors that include the number of transactions and their monetary value. Currently, those parameters are rules-based. 

Visa will soon introduce an upgrade to its stand-in processing service called Visa Smarter STIP. The improvement incorporates artificial intelligence (AI) models to add flexibility to the strict rules-based decision engine deployed by the existing STIP service. 

The new model closely mimics the decisions an issuer’s system would make when it receives an authorization request.

To create Smarter STIP, Visa took two years of card authorization data, billions of records, and separated it into two tranches. One was used to train a deep learning model. The second tranche was used to test the accuracy of that AI model.

Visa says that on average, the model delivered authorizations in sync with an issuer’s in-house capability 95% of the time. In a Smarter STIP simulation with a participating issuer, as many as 50% of transactions that would have been declined using only rules-based decisioning could be authorized—and without any increase in fraud. 

Visa Smarter STIP uses AI to improve the authorization step of transaction processing. Uses for AI in the clearing and settlement steps are imminent as are other services related to transaction processing. Issuers can sign up for Visa Smarter STIP beginning this October. 


Gourab Basu is VP Global Network Product and Strategy at Visa in San Francisco, California,,

© Copyright 2020 Nilson Report

You have 1 free articles remaining. Subscribe today. View Subscription Offer
New subscribers receive over 130 articles in the 22 issues published each year,
plus the last five years of issues (that's over 1,200 articles) on a searchable flash drive.