The US data appear to be based on this Federal Reserve info (or a similar document from a previous year): https://www.frbservices.org/binaries/content/assets/crsocms/... . However, that data includes both in-person and remote payments. The proportion of cash for in-person is slightly higher. There is also a table showing "in-person payments less than $25", and in that category cash is about 29%. This aligns with my own experience that people are fairly likely to use cash for small transactions, but cash becomes more and more unlikely the larger the transaction.