Every year, hundreds of new image analysis tools are created and released for analysis of microscopy data. Few of these will stand the test of time though - with many becoming unusable within a few years (or even months!) of publication. How can we make long-term maintenance and update of these tools easier? How can we ensure new updates don't break existing functionality? In this talk, I will focus on 'good software practices' that help make your software more robust e.g. automated testing, dependency updates and versioning. I will give examples from Fiji / napari plugins (as these are very popular for volume electron microscopy data), but the general principles should be applicable to any kind of image analysis software.