Thoughts are not crimes in free countries. What someone "might" do is irrelevant.

Amazon thought that the person had already shouted a slur at the driver. They were wrong, but if they had been right blocking further deliveries to protect the drivers would have been reasonable (if it was just that)

I disagree. It should take something like a police report to block further deliveries. Just taking one delivery driver’s word is ripe for abuse.

So Amazon should continue sending people to someone who will racially abuse them just because the police didn’t catch it yet? (again, assuming that they had done the investigation properly which they didn’t in this case, which would involve collecting more evidence to back it up)

I said a police report, meaning the driver would report it to the police. Not that they'd catch them in the act.

> someone who might racially abuse them

Agreed, but note that "abuse" is the operative word here. It's possible to be racist and not be outwardly hateful, and it's also possible to be non-racist and horribly abusive to staff.