Of course you do, when you start a project you know how you're going to do it. Even if you never done it before and it might fail, you know what you're going to try, so you know from the outset if it's patentable or if your attempt is based on someone else's approach so it's not patentable.
> when you start a project you know how you're going to do it
In my experience that has only been true for the simplest tasks I've been assigned to do. The only examples that come to mind are stuff I did as a junior developer. YMMV, of course.
As my career progressed more towards the R part of R&D, uncertainty skyrocketed.
