Both anecdotally and empirically, it is a familiar observation that many students are not proficient in basic skills used in STEM courses. These fundamental skills often incur significant cognitive load for students, potentially inhibiting complex problem solving and deep thinking about relevant concepts. A natural reaction to such a deficit is to help students improve proficiency in such skills. Often, instructors present or assign a brief practice of the skill; then it is assumed, frequently incorrectly, to be mastered. There are also online learning products on the market to help practice some skills. To study and improve on such practice, we employed a general theoretical framework and empirical research to iteratively design and implement “STEM Fluency”, an online practice application aimed at maximizing fluency (accuracy + speed) and retention to reduce cognitive load. For some skills relevant to introductory physics, we have documented that significant improvements in fluency can be relatively efficiently achieved. As such, a principled and research-based framework for building content and structure will be discussed with an eye towards facilitating participant reflection on applications relevant to their context. However, this issue is not so straightforward and important questions remain. For example, how can basic skills be practiced to enhance deeper instructional goals such as conceptual understanding and complex problem solving? What should we do about basic skills that appear unmalleable in a one-semester course? Which basic STEM skills will be useful beyond STEM courses and so should be mastered? Investigating these questions seems critical to meaningful progress.