All software relies on data, from the lines of code that ‘make stuff work’ to system adaptation parameters from real world sensors. As engineers we seek to build a system that provides ever-more complex functions, adding features to give our products the ‘edge’. Increasingly, these features are reliant on data. Data in software is assessed as an integral part of functionality design, but this does not cover everything. Data has properties which can be lost or modified, leading to a situation where data (not the software function) contributes to harm. Changes or errors in the data can cause software to behave in an unexpected way, defying the outcomes it was carefully designed to achieve. This presentation fills the gap by describing the Data Safety Guidance methodology for managing the risks associated with data, specifically in the software development life cycle.