Currently, it's by far easier for everyone to use Visa: most people have access to that system, services are widely available for businesses of all sizes, etc. The primary drawbacks are transaction costs and businesses being on the hook for most fraudulent transactions.
Switching to Ethereum would make things slower, but that could improve. Since it's not tied to a hard currency, however, there's a much bigger problem cost management: I know exactly how much a Visa transaction will cost but gas fees are unpredictable and volatile, and the exchange rates for ETH vary both independently and more than most currencies. Similarly, there's a real appeal to not automatically being on the hook for a disputed transaction but that's significantly undercut if you have to worry about being one mistake away from irrecoverably losing everything in your account and so far customers have not been jumping to take on more personal risk either.
The big question here is also what competitive businesses do in response: for example, if at some point in the future Ethereum actually became cost-competitive what happens when Visa simply lowers their transaction fees until that's no longer the case again? They have a lot of margin to do that and the merchants don't need to change anything about how they do business. Similarly, Ethereum is nowhere near the speed of a traditional credit card transaction but even if it hit that speed it'd be playing catch-up with Apple/Google Pay – businesses care a lot about things like that since it's often highly visible to customers and can affect things like retail lines, so the question is whether that can be improved faster than companies like Apple/Google, Stripe, Square, etc. can improve their services.