As Mark commented above, Soft Operations Research is fully part of Operations Research. Soft OR focuses on approaches and methodologies to state and model the complex problems encountered in the real life of organizations. In contrast, "Hard OR" refers more to the mathematical and algorithmic foundations of the discipline, that is, how to solve the problem once it has been stated and modeled.
A classical sentence about modeling is that this is more of an art than a science. Nevertheless, having a scientific and systematic approach to the definition, structuring, and modeling of real-life problems is critical for the advance of OR and for its success to improve society. Like any other field which is at the same time a practice, a technology, and a scientific discipline.
In particular, Soft OR raises one interesting point: this is critical to well understand the people for which we build an optimization model or an optimization software solution, to solve a problem. Not only to define the problem but also to understand why they wish to solve this problem (that is, the context around the problem), and in the end, to ensure that our work and our solution will fully satisfy their needs, beyond the problem. Indeed, answering the needs of our clients and users is the only way to create value for people and society. Now, answering the needs is not only a matter of maths and algorithms. All of us - OR scientists, teachers, and practitioners - we must always remember this point if we wish our discipline to be successful in practice.
For an introduction to Soft OR, we encourage you to read this article published in the INFORMS Analytics magazine or this JORS paper. The J.E. Beasley notes or this webpage of the Military Operations Research Society (MORS) are also of interest.