Skip to content

Case study · Plant operations platform · 2026

Durus Running a galvanizing plant on one screen

Rebuilding the order-to-invoice workflow of a hot-dip galvanizing plant — turning five disconnected paper-and-spreadsheet steps into one desktop system that tracks every order from intake to the plating line to the invoice.

Role
Product Designer & Team Lead
Team
3 designers · 1 dev · 1 PM
Timeline
5 months
Platform
Desktop Application
Durus — the eight-stage processing board showing the plating line in real time

01

Overview

Project info

Durus is an internal desktop application built for a hot-dip galvanizing plant — the factory process that coats steel parts in zinc to stop them rusting. It runs the full order lifecycle: customer orders, incoming-goods check-in, QC “black inspection,” the eight-stage plating line, checkout and invoicing. I led design for a ground-up redesign of the core operational surfaces, owning both the hands-on work and the design direction for a cross-functional team of 3 designers (I led), 1 developer and 1 PM, working alongside planners, QC inspectors and floor operators who use the app every shift.

71% → 93% On-time order completion
12% → 4% Re-dip rate — batches needing rework after QC
−58% Time to schedule & release an order to the plating line

02

The challenge

The plant ran a real, physical production line, but the software tracking it didn’t reflect how galvanizing actually works. Orders, jigs and quality checks lived in separate lists with no shared state — so a planner couldn’t see where a job sat on the line, and a re-dip meant someone finding out by walking the floor.

Invisible process
The plating line has eight physical stages — hanging, degreasing, pickling, galvanising, dispatching and more — but the software only showed “in progress” or “done.” No one could see a jig’s real position without walking out to look.
No domain logic
Recipes, gauge, size and hangability determine how a part is actually dipped, but intake was a bare form — the same screen for a light bracket and a heavy structural beam.
Disconnected records
Customer pricing, order history and QC evidence lived in different places, so planners and finance re-asked the same questions on every call.

03

The before

What the core plant-floor screens looked like before the redesign — flat lists with no process tracking, no filtering, no domain logic, and no evidence trail.

Legacy processing — one undifferentiated job list with no stages or jig status
Processing — one undifferentiated job list. No stages, no jig status, no priority, no read on where anything sits on the line.
Legacy order schedule — a bare order list with no filters, status pipeline or actions
Order Schedule — a bare order list. No filters, no status pipeline, no priority, no actions, no paging.
Legacy customer — a read-only contact card with no linked orders or pricing
Customer — a read-only contact card. No linked orders, no pricing, no notes, no way to act from the record.
Legacy black inspection — a skeletal intake form with no recipe or evidence
Black Inspection — a skeletal intake form. No recipe, no hangability check, no photo evidence, no approval step.

04

Approach

  1. Research

    Phase 01 · research Discovery

    Spent time on the plant floor with QC inspectors, planners and line operators, and walked the physical galvanizing line stage by stage. Found the same job being tracked three different ways — a paper traveller on the floor, a spreadsheet in the office, and a memory in someone’s head.

  2. Information architecture

    Phase 02 · information architecture Structure

    plating stages modelled into one order record

    We re-modelled the order as the spine of the whole system, not a row in a spreadsheet. An order now carries its customer, its materials and recipes, its position across the eight-stage plating line, and its invoice — so a stage change on the shop floor, a QC approval and a finance update all move the same record instead of three disconnected ones.

  3. Systems & leadership

    Phase 03 · systems & leadership Scale

    As team lead, I set the direction for three designers and built a shared component system (stage cards, status pipelines, filter bars, evidence panels) so development shipped consistently across intake, processing, QC and invoicing without the app fragmenting screen to screen.

  4. Prototyping & validation

    Phase 04 · prototyping & validation Proof

    Tested the Processing board and the Black Inspection flow with real operators and inspectors on the floor; iterated the jig-readiness indicator and the photo-evidence step until both felt fast enough to use mid-shift, not just correct on paper.

We stopped tracking paperwork about the process and started modelling the process itself.

05

The solution

A processing board that matches the real line

An eight-stage Kanban — hanging through dispatching — with live jig-readiness, recipe grouping and priority bands turns a flat job list into a real-time picture of the plating line, so a planner can see exactly where every dip sits without walking the floor.

Processing board — an eight-stage Kanban with live jig-readiness and priority bands
Processing board, real-time — final screen.

An order schedule planners can act on

Filters, a colour status pipeline and per-row actions turn a static order list into something a planner operates — check status, adjust priority and release a job to the line without leaving the table.

Order schedule with filters, a colour status pipeline and per-row actions
Order schedule, operable — final screen.

A customer hub, not a contact card

Contacts, recent orders, zinc-rate offers and notes now live on one record, so a planner or account handler answers “what’s this customer’s history and pricing?” in one screen instead of three.

Customer hub with contacts, recent orders, zinc-rate offers and notes on one record
Customer hub, connected — final screen.

QC that understands the process

Recipe linkage, a hangability check, photo evidence and an approve step turn black inspection from a data-entry stub into a domain-aware QC workflow with a defensible audit trail — built around how galvanizing QC actually works, not a generic form.

Black inspection with recipe linkage, a hangability check, photo evidence and approval
Black inspection, with evidence — final screen.

06

Results

71% → 93% On-time order completion
12% → 4% Re-dip rate — batches needing re-galvanizing after QC
−58% Time to schedule & release an order to the plating line
−45% Time to complete a black inspection, with photo evidence & audit trail
−40% Time from checkout to invoice
+22% Plating-line and jig utilization

Within two quarters, the plant stopped running on paper travellers and started running on one system. Planners could see a job’s real position on the line instead of asking the floor, QC inspectors closed a black inspection with evidence attached in a fraction of the time, and finance moved from checkout to invoice in days instead of the better part of a week — while the line itself ran fuller, with fewer idle jigs between stages.

07

Reflection

The hardest part wasn’t designing a better-looking form — it was designing the plant’s real workflow, stage by stage, so the software matched what actually happens to a piece of steel between intake and dispatch. The order and process data model was the real unlock: once a job’s recipe, stage and QC status lived on one record, every screen — the board, the schedule, the inspection, the invoice — had something true to show. If I ran it again, I’d get operators and inspectors into the information-architecture phase even earlier; the stage model only held together because the people running the line agreed on it before we drew a single screen.

Easycompliance — the connected GRC risk register

Restart Showcase

Easycompliance / Risk management

Continue

08

Let’s Connect

Looking for a designer who can bring clarity to complex products and improve business outcomes? Let’s connect.

Download CV