01 / Logistics
Rebuilding freight visibility for 400 ops users
Discovery interviews, IA rewrite, and a component system shipped in 12 weeks.
Context
Logistics / freight forwarding
Mid-size 3PL operator, ~1,200 employees
Company name withheld under NDA.
Role
Me — Discovery research, information architecture, interaction design, component system
Team — 1 product manager, 3 engineers
The constraint
400 ops staff were tracking shipment status across four disconnected legacy tools, phoning each other to confirm what the software wouldn't show. Leadership wanted a fix in one quarter, with zero downtime during peak shipping season — the one window where the old tools genuinely couldn't be touched.
What we tried
01
Tried — Design a single dashboard to replace all four tools in one release.
Learned — In testing, ops staff couldn't verify the new system against workflows they already trusted — especially mid-peak-season, when a wrong number costs real money. Adoption stalled before it started.
Did — Shipped a read-only visibility layer that ran alongside the old tools first, so staff could cross-check it against systems they already trusted before any workflow depended on it.
02
Tried — Carry over the existing status codes from the legacy systems unchanged, to keep migration simple.
Learned — The status codes were themselves the root confusion — three systems used the same code to mean different things. Carrying them over would have just relocated the bug.
Did — Ran a card sort with ops leads to build one shared status vocabulary, then mapped every legacy code onto it explicitly, surfacing conflicts as flagged exceptions instead of silently picking one.
03
Tried — Migrate all four tools' write-actions (updating status, logging exceptions) in the same release as visibility.
Learned — Bundling write-actions with visibility meant one bug could block adoption of both — too much risk for a single release during peak season.
Did — Split the rollout: visibility first, write-actions migrated tool-by-tool over the following two quarters, each with its own rollback path.
Outcome
-42%
Status-check calls into ops dropped 42% in the first quarter after the visibility layer shipped, and the full write-action migration completed on schedule with no unplanned downtime during peak season.
