Three directions
Each card has a Preview this direction button. It re-themes the entire prototype — every screen, until you switch back — because all three are the same tokens with different values. Walk the till and a stock table in B before deciding; a palette judged from a swatch row is a palette judged in the wrong place.
Painted-concrete ground, ink for everything that acts, and one hi-vis marker that only ever points — never writes. The till and every commit gate are a graphite panel set into the desk, like an instrument bolted into a workbench.
Hi-vis is the vernacular of the work itself: vests, floor tape, forklift stripes. It marks position, which is what a multi-tier system is about.
Dark throughout, not just at the till. A loading bay at 05:00 and a shop counter at 22:00 are both dim, and a bright screen in a dim room is the thing people actually complain about. Data glows; chrome recedes.
A cool teal signal has the contrast on dark that hi-vis has nowhere else, so it can carry type — which frees up the whole accent rule.
Each tier owns a hue and the whole screen takes it: warehouse blue, store violet, counter orange. Most staff only ever work one tier, so they learn one coloured world and know instantly when they are looking at someone else's.
It is the most immediately legible of the three on a shared screen across a room, and the easiest to explain to a new hire in one sentence.
All three work. A wins on the thing this product does most: reading a number and trusting it. Its palette spends nothing on decoration, so green, amber and red keep their full meaning on a stock table, and the one accent goes entirely to saying you are here and this changed — which is the hard problem in a system with three tiers and eleven locations. B is the right answer if most screens are actually read in dim rooms; C is the right answer if staff turnover is high and training time is the real constraint. Say which of those is true and the direction should change.