DataWars is built around a problem that anyone who has tried to learn data science will recognise instantly, which is that you can finish course after course, watch every video, and still freeze the moment someone hands you a real dataset and asks you to actually do something with it, because watching someone else clean data and build a model is a completely different skill from doing it yourself. This gap, often called tutorial hell, is where an enormous number of learners get stuck, and DataWars sets out to fix it not by teaching more theory but by giving you a large library of hands on projects and labs to practise on. The projects are based on real, messy datasets rather than the pristine toy examples that theory courses tend to use, which matters because the actual craft of data work is largely in wrangling awkward, incomplete, inconsistent data into something usable, and you only get good at that by doing it repeatedly. A genuinely nice touch is that everything runs in the browser through hosted notebooks with no local setup, so there is no friction of installing Python and fighting with environments before you can start, you just open a lab and get straight into working with pandas, SQL, and the rest.
I see it as a strong complement to the interview practice sites, because where those sharpen your querying and algorithmic reflexes, DataWars gives you end to end applied projects that look more like the actual job, filling the space between abstract theory and real work. The caveats are mostly about knowing what it is for. DataWars is a practice ground, not a from scratch teacher, so it assumes you already have the fundamentals and want reps rather than first explanations, and if you arrive without knowing the basics of Python or statistics you will struggle and should learn those elsewhere first. Getting real value out of it means a subscription, since the free tier only stretches so far, and its centre of gravity is data science and analytics rather than deep machine learning theory or modern AI engineering, so it is not the place to learn to build large language model applications.
As with any self directed practice, it also demands discipline, because there is no cohort or deadline nudging you forward, and the people who benefit are the ones who actually put in the hours. My recommendation is a clear yes for a specific situation. If you have worked through the courses, feel like you understand the ideas, but seize up when faced with a real problem, DataWars is aimed squarely at you, and the honest way to think about it is as the gym where you convert passive knowledge into genuine skill through repeated practice on realistic problems. Pair it with interview prep if you are chasing a data role, bring your fundamentals with you rather than expecting it to supply them, and put in consistent reps, and it does a good job of bridging the stubborn gap between having learned data science and actually being able to do it.