One thousand grids
Generating a thousand column grids to find out how few actually differ.
We generated a thousand column grids, printed them at A3, pinned them across the studio wall and spent a Friday crossing out the ones nobody would ever use.
Eleven survived. The interesting part was not which ones — twelve columns, unsurprisingly, does most of the work — but why the others died. Nearly all of them failed on the same thing: they could not hold a two-column body text and a full-bleed image without one of the two looking like a mistake.
The interesting part was not which ones — twelve columns, unsurprisingly, does most of the work — but why the others died.
The exercise sounds like procrastination and to be fair, partly it was. But it produced a defensible default. When a client asks why the site is on twelve columns, the answer is not "it is standard", it is "we tried the alternatives at full size and these are the eleven that survived contact with real content".
The wall stayed up for a month. Two of the discarded grids came back for a specimen book where the rules were different, which is the other lesson: nothing is wrong, things are wrong for a job.