Reasoned discourse about software requires that identity conditions of a software product preserve its identity over updates. Despite its failure to produce such identity conditions, the view that they can be formulated in terms of algorithms or code is nevertheless believed to be on the right track. In this talk, I shall argue to the contrary, as the resultant identity conditions cannot be evaluated consistently by independent parties. Instead, approaching the problem from a user’s perspective, I arrive at a necessary condition for the identity of software products. This condition induces what I call use-plan-compatible equivalence classes. I show how their associated identity conditions both preserve identity over updates and can be consistently and independently evaluated. Still, not based on a sufficient condition, the concept of a use-plan-compatible equivalence class is arguably too broad. Thus, the results call for further discussion of how to delimit more accurately those objects that should fall under the concept of a software product.