Observability

This chapter explores observability as a core discipline in growth engineering. Growth without observability is guesswork. The concept of observability comes from control theory, where it was formally defined in the 1960s by Rudolf E. Kalman, a pioneer in modern control systems. Observability also powers experimentation. It allows us to instrument which users saw which product variants, and helps us analyze the data to determine which variant performed best and is ready for productization. Observability is made possible by instrumentation and telemetry. Instrumentation is the first step to turning a product into a feedback machine. Telemetry is the data emitted from instrumented code. The chapter presents some common anti-patterns that can reduce the value of observability setup, along with ways to avoid them. Following is a breakdown of commonly used tools across the three main layers of observability: instrumentation, telemetry collection, and visualization and analysis.