Data Engineers: What If Your BigQuery Function Could Return Multiple Tables?

When working with Google BigQuery, one limitation often catches many engineers off guard: A table function can return only a single table with a predefined schema. However, in real-world data pipelines, the output is rarely just one dataset. A typical pipeline may generate several types of results, such as: Cleaned or transformed data Validation errors Pipeline metrics Execution logs This raises an important question: What should you do if your function logically produces multiple outputs? Many engineers assume the only option is to use multiple queries or stored procedures to handle each output separately. But there’s actually a clever workaround that allows you to structure multiple outputs within a single function. Let’s take a closer look at how this can be achieved. 🚀 The Problem Suppose you are building a data pipeline that processes raw data and produces: Final transformed data Validation errors Pipeline execution metrics Ideally, a function would return multiple result sets. Ho