August 2026 · 6 min
Most design system arguments are jurisdictional
Two designers are arguing about a button colour. Both of them are right. This is the most common meeting in design-systems work, and almost nobody names what's actually happening in it.
The argument looks like taste. It isn't. It's a border dispute.
Why the argument never resolves
One designer is protecting brand coherence. The brand has a signature colour, the product should feel like the brand, primary actions are the most visible thing on the page, so primary actions should carry the brand colour. Every step of that is sound.
The other is protecting usability. The interface has a language where colour means state. Colouring an action by brand rather than by function breaks that language, and in a dense product breaking it is expensive. Also sound.
The meeting runs long because both people are answering what do we prefer, and preference has no tiebreak. The question that has an answer is which layer decides this.
The fix is a written order
One section of a governing document. Five items, ranked:
That's it. It isn't clever. Its value is entirely in being written down before the argument rather than improvised during it.
Now the button conversation takes ninety seconds. Brand coherence sits below usability. The brand colour loses. Nobody's judgement was overruled. The order was applied, and everyone in the room agreed to it in advance.
The clause that makes it usable
A priority order on its own can feel like a demotion of brand, which is how you lose the brand team's support for the whole system. So the order needs a companion rule:
When a brand expression weakens usability, translate the brand's intent rather than copying its visual treatment.
This is the difference between governance and suppression. The brand wanted the product to feel confident and distinctive. Copying the hex was one way there, and it happened to be the expensive one. Translating the intent through density, typography, motion and shape gets the same outcome without spending the colour system to do it.
The brand team isn't being told no. They're being told not with that particular instrument.
Where the order has to live
A priority order buried in a slide deck does nothing. It has to sit in a document people cite, not one they half-remember. Three properties make that work.
It sits above any single project. If the order lives in a design system's readme, it applies to the design system. If it lives in a constitution above it, it applies to the pattern layer, the internal studio and the brand team too.
It states ownership boundaries. For each layer: what it decides, and what it doesn't. Most jurisdictional arguments are really about undeclared boundaries, and the fastest fix is a table with a does not own column.
It has an inheritance rule. A lower layer may specialise a higher layer, but must never silently contradict it. The word doing the work is silently. Contradiction is how systems evolve. Contradiction without a marked, reasoned deviation is how they rot.
What you actually get
The order doesn't make better decisions. It makes decisions cheaper, and cheapness compounds. A team that settles a boundary question in one meeting instead of three has meaningfully more capacity, and the difference shows up as speed nobody can attribute to anything.
The second effect is better. Once the order exists, people stop bringing the argument. They read it, apply it themselves, and only escalate the genuinely new cases. The document's success metric is how rarely it needs to be invoked.
The catch
A priority order with no record of use is a PDF. Until you can point at a specific dispute it settled, with names and a date, you've written an intention, not a governance mechanism. The first real citation is worth more than the next five sections.