Accurate and up-to-date mapping products often form an essential part of management, conservation and restoration activities in marine and coastal ecosystems, particularly key habitat such as seagrass, mangroves, saltmarsh, mudflats and coral reefs. However, producing detailed, fit-for-purpose mapping products for these ecosystems can be a challenging and time-consuming task, particularly when end users of each application might require a different combination of the wide range of satellite image data sources, environmental data layers and training/validation data. To address this challenge, we developed a flexible mapping workflow that can be customised to provide a bottom-up mapping solution in a range of coastal environments. The framework can map different coastal ecosystems simultaneously from any commonly available satellite imagery and environmental data, using a range of common field data types for training and validation. We demonstrate the method on Landsat 8 (30 m), Sentinel-2 (10 m) and Planet (3 m) satellite image data, mapping the key coastal ecosystem types (seagrass, mangroves, saltmarsh, tidal flats, and coral reefs) using both geo-tagged field data and expert annotated training data sets. Examples are demonstrated for several ongoing projects within Australia relating to environmental ecosystem accounting, restoration, conservation planning and environmental monitoring.