How do I tell which suppliers are actually risky before one fails?

Right now you find out a supplier was risky the hard way, when they fail. That is a painful and expensive way to learn, and wanting to see it coming instead is exactly the right instinct. The skill you are missing is not luck. It is a way of looking that you can build. Risk is not evenly spread. A supplier is dangerous to you in proportion to two things: how much damage their failure would do, and how likely that failure is. The ones to watch are where both are high, the supplier you cannot easily replace who also shows warning signs. Most suppliers are not real risks. A few are, and those deserve nearly all your attention. Make a simple list of your suppliers and mark, for each, how badly it would hurt if they vanished and whether you have any backup. The ones that would hurt badly with no backup are your real risks, before anything goes wrong. Start watching those closely and finding second sources. You do not need to predict every failure. You need to know where you are exposed.