STEM educators have taken on a new and radically different reality: that students will have access to online tools and online knowledge as they undergo assessment. Students may look up facts that are easily available online, employ online tools and also employ paid services to cheat. The first wave of novel assessment strategies devised in response to this new reality was online invigilation measures that were often either ineffective or widely hated by students. This paper considers two frameworks for effective online assessment in the modern era: a framework for question design in STEM and a framework for the structure of assessment. The framework for question design discusses methods to discourage the use of online tools and calculators to simply obtain answers and methods to detect contract cheating. The framework for assessment structure discourages cheating while managing to increase student satisfaction and engagement with content: we present a case study from a cohort of 113 first-year university students of adapted assessment strategies with no identified paid cheating, well-differentiated student results and large-scale positive student feedback. Our key contribution is to identify simple and effective steps to assess students fairly in an online environment, without empowering or motivating them to cheat.