It is the nature of startups that most will fail. Startups are hard. They fail for many reasons: wrong product, wrong price, wrong distribution channel, weak marketing, and so on. Part of what we teach in Lean Startup (also known as Evidence Based Entrepreneurship) is how to reduce the risk of failure. Statistics have shown (for years) that more than 90% of all startups will fail.
With systematic and comprehensive customer discovery we can significantly reduce the risk of failure.
But startups will still fail for reasons unrelated to product or market. Customer discovery is little help against changing regulations, new competitive forces or a founding team that implodes.
So, we should teach student founders identify failure modes. The faster the founders can diagnose a problem and respond with a course correction, the more likely they are to survive.