We rebuilt trade assignment three times. The first two versions were thorough, configurable, and correct. Nobody used them. This is the story of the third version, which works because it asks almost nothing of the person using it.
The 90-second constraint
The design brief came from watching supers, not from a feature request. A site supervisor moving through a building does not stop to "do data entry." They capture a defect in the gap between two other things — a contractor waiting, a phone ringing, a delivery at the gate. The realistic budget for assigning a defect to a trade is about 90 seconds, and most of that is walking.
So the question stopped being "what fields should assignment have?" and became "what would a super do if they had 90 seconds and one thumb?"
Two taps, no thinking
The answer we landed on is two taps:
- 01Tap the defect's trade — tiling, painting, electrical, waterproofing.
- 02Tap the contractor.
That's it. The defect is assigned, the contractor is notified, and the super has moved on. No form, no modal stack, no required fields hiding behind a "more options" disclosure.
The trick is that everything else assignment could capture is either inferred or deferred:
- →The location is already on the defect — it was captured when the photo was taken.
- →The due date defaults from the project's workflow rules, not a date picker.
- →The notification to the contractor fires automatically; the super doesn't compose it.
- →The detail the office cares about gets added later, at a desk, by someone who isn't standing in a stairwell.
Defaults are a design decision, not a fallback
The reason most assignment flows are slow is that they treat every field as a question the user must answer. We treat every field as a default the system should already know, and an exception the user can override if they need to.
A contractor's default trades, a project's default close-out window, the standard notification template — these aren't settings buried three menus deep. They're the reason the common case is two taps and the rare case is still possible.
The best field interaction is the one that's already correct before you touch it.
What we deliberately left out
Every version of this we rejected had more power and less use. We left out per-assignment priority pickers, free-text instructions at point of assignment, multi-trade splits, and a dozen other things that sound reasonable in a planning doc and die on contact with a wet Tuesday on site.
None of them are gone — they live in the defect's full detail view, available when someone has the time and the keyboard for them. They're just not in the critical path, because the critical path belongs to the person with 90 seconds and one thumb.
That's the two-tap standard. Not "fewer features." Fewer decisions, in the one place where decisions are expensive.