As the volume and variety of data grows, in parallel with organizational investments in analytics and machine learning, ensuring data is high quality has become mission critical for many organizations. But traditional software monitoring and remediation methods tend to fall short here, which has given rise to a set of approaches called data observability. We’ll explore the elements of a general data observability strategy, pitfalls to avoid, and nuances around data observability for machine learning applications. We’ll draw on real-world lessons from Opendoor and other leading ML companies, ranging across industries and touching on structured and unstructured data.