Great article! Another one to ask is how they fix mistakes — do they have blameless post mortems for example?

Do they use tests? Do they have human testers? How do they decide when to use software tests?