Over the past two years as a founder, I’ve learned that in building successful products, perfection is only necessary where it matters.

Every task falls into one of two buckets:

• Mission-critical problems: These demand 100% accuracy and anything less than perfect is unacceptable. Think autonomous driving, medical diagnoses, or banking transactions.
• Non-critical problems: For these, around 80% accuracy is good enough, 90% is amazing. Examples of non-critical problems are googling for a chicken parmesan recipe or writing a blog post.

Building a perfect solution for non-critical tasks is a waste of time. At the same time, trying to get 99% accuracy on a mission-critical task is a fool’s errand. We wasted a ton of time building a 99% accurate solution for a client who needed 100%—and they walked away.

Focus your effort on perfecting only mission critical things, and get everything else good enough to iterate later.