Every project at Knurl starts with a spreadsheet, not a sketch. Before anyone draws, we take apart the product being replaced, or its closest competitor, and list every part: mouldings, screws, clips, labels, cable ties, all of it. The predecessor of the Halo desk lamp came to 41 parts. Eleven of them were fasteners, in four different sizes.
What the count tells you
A part count is a rough proxy for almost everything a client cares about. Each part needs a tool or a supplier, a drawing, a tolerance, an incoming inspection, a bin on the assembly line and a line in the service manual. Each one is also a place for the product to rattle, leak or fail. When we plot our own past projects, assembly time follows part count more closely than it follows size or price.
Does it move relative to its neighbour? If not, the two can probably become one.
Does it have to be a different material? If not, same answer.
Does it need to come off for repair? If yes, it stays separate and gets a standard fastener.

The old lamp laid out on the bench: 41 parts, 11 of them screws.
The cheapest part is the one you never tool.
Those three questions are blunt on purpose. On Halo they took the arm from six parts to two, an extrusion and a cap, and replaced four screw sizes with one. The lamp ended at 17 parts and opens with a single screw. Counting is not the goal, though. A product with 12 parts that cannot be opened is worse than one with 20 that can, so the count sits next to the repair score on the wall, and the two argue until the drawing starts.

Written by
Ravi Menon
Partner, engineering
Keeps a spreadsheet of every screw we have ever specified and wants it shorter.


