There’s a moment in every CAD project, about three months in, where the real cost of the early design decisions becomes visible. It’s the moment the client comes back and says: “Actually, can we make it 12 mm wider, and can the mounting pattern change to match the new frame?”
If the assembly was built with parametric discipline, this is a ten-minute change. If it wasn’t, it’s a week of rework. Same client, same scope, same model, the difference is entirely in how the first fortnight of CAD work was approached.
What parametric actually means
In practice, parametric design means every dimension that could conceivably need to change is driven from a single source of truth, usually a design parameter table at the top of the assembly tree. Every downstream feature references that parameter. Nothing is hard-coded. Nothing is drawn in a sketch and then dimensioned by dragging.
Most of us learn CAD by clicking-and-dragging. It feels faster. For a single bracket that nobody will ever revisit, it is faster. For an assembly that will survive three client revisions and a manufacturing run, it’s catastrophically slower, the rework cost compounds.
The commercial case
This isn’t a craft argument, it’s a commercial one. RKT’s quote for a project is informed by how many revisions we expect, and how painful each revision will be. Assemblies built with parametric discipline are cheaper to quote, cheaper to deliver, and cheaper to maintain.
On a typical yacht-hardware engagement, we budget about 15% of our CAD time in the first fortnight on parameter-table construction and master-sketch work. That 15% saves us about 40% of the total project time over the full engagement lifecycle, because every downstream change runs through rebuild instead of rework.
Where practitioners go wrong
The most common failure mode is building the parameter table retroactively, trying to parameterise an assembly that was already built click-and-drag. This almost never works. The brittleness is baked in.
The second most common failure is over-parameterisation: making every single dimension driven, even ones that will never change. This slows the model down without delivering downstream benefit. The craft is in choosing which parameters matter.
The heuristic
A useful rule: any dimension that appears on a drawing the client will see, or any dimension that references another part’s geometry, goes in the parameter table. Everything else can be a sketch dimension.
Follow that rule, maintain a naming convention for parameters, and structure the assembly tree so that the parameters sit at the top, and the next three revisions will cost you hours rather than weeks.
