Enterprise workflow design for a 24/7 railyard
Norfolk Southern’s shipping clerks ran freight orders on a decades-old green-screen mainframe. Over a Fjord engagement, I led the design of a cloud-native concept to replace it: fast enough for clerks working full shifts, and clear enough to finally give supervisors a live view of the yard.
Role
UX Lead
Client
Norfolk Southern
Agency
Fjord
Platform
Enterprise web
When
2019 · cloud-native concept
Today’s Shipments · Priority queue · Live status Terminal operating system · Web
The 60-second version
Norfolk Southern runs one of North America’s largest freight networks. Its shipping clerks managed order processing on a mainframe that demanded hundreds of memorized command codes. The job wasn’t to reskin it. It was to answer a harder question: can you modernize a system that can’t go down, for people who already know it cold?
The green screen worked if you’d put in the years: dense fields with no hierarchy, ASCII navigation, function keys worn into muscle memory. It was fast in expert hands and merciless to everyone else, and the learning curve kept the knowledge locked in a handful of veterans.
Supervisors and executives needed something the terminal never gave them: a live read on order health, exceptions, and trends, not a report assembled by hand after the shift. Clerks needed the same speed without the memorized codes, and both had to live in one system.
Command-line terminals and function-key navigation: the operational backbone of a major railway, built for another era of computing.
Speed up order processing for clerks and give supervisors and executives a live operational view, from the same product, not a second reporting stack.
The insight that drove every screen
Green screens stored freight as flat text and command sequences. Workers thought in physical reality: containers, exceptions, priorities, customers, time. The design job was to translate that ontology into a spatial interface, so clerks could act in the moment and leadership could see the aggregate without waiting on a spreadsheet.
1Make work clear
The legacy failure wasn’t missing data. It was data without a primary job on each screen. Every screen needed a clear action path, with secondary detail available but not in your face.
Instead of packing every field into one terminal page and calling that ‘complete.’
2Earn trust
These weren’t casual users. Unionized clerks spent full 8-hour shifts in the terminal, many with years or decades of muscle memory in it. They’d be trained on whatever replaced it, so the bar wasn’t ‘easy for a first-timer,’ it was ‘faster than what they already do without thinking,’ with the training cut way down.
Instead of a modern coat of paint that ignores how much expertise already lives in the room.
Discovery meant bridging the tower and the boardroom. Clerks needed speed under pressure. Supervisors needed signal without waiting on manual reports. Basic questions stayed open: What’s blocked right now? and Who can see it in time?
Landscape audit
We studied the current terminal state in the field, and looked outside rail to SaaS products that already help knowledge workers prioritize and clear high-volume work. Three gaps kept showing up.
Terminals buried priority inside command sequences. Nothing felt like a triage queue.
Leadership reconstructed KPIs from delayed reports. Live exceptions stayed invisible upstairs.
Hundreds of codes made competence a multi-year investment. New hires paid for the system’s age.
The current state · legacy terminals
How might we give terminal workers real-time visibility into container logistics, and cut the cognitive load of managing complex freight transfers?
Status quo vs. what we designed
Those gaps set the contrast: replace cryptic command density with scannable work, and give supervisors a live read without a second reporting stack.
Clerks navigate by memory. Supervisors wait on assembled reports. The yard’s state exists, but not as a shared, timely picture.
Clerks clear prioritized work on modern screens. Executives watch KPIs and exceptions while operations are still in motion.
Two mindsets
Research required different methods for each group, because they needed different truths from the same platform.
Target A
Target B
Into the work environment
Clerks were observed on live terminals under real conditions. Supervisors and executives sat for structured interviews about the information they needed and could not get.
These clerks had operated green screens for decades. We had to earn their trust by making the job genuinely easier.
Previous pattern · Mainframe terminals · Institutional knowledge as gatekeeping
Other railway tools shared the same legacy problems. We looked outside the industry to SaaS products that already help people manage, prioritize, and finish high-volume work with clarity.
Zendesk-style queues showed how to treat flagged shipments like tickets: status, priority, and a clear next action.
Relationship history in one view translated into a shipment lifecycle in one place, not scattered across terminal screens.
Priority and recency at a glance informed Today’s Shipments: what needs attention now versus what can wait.
Design principles
Four rules, drawn from constraints and research, revisited at every review so polish never outran the mandate.
Principle 1
Every screen has a primary job. Secondary information stays accessible without competing for attention.
Principle 2
Zero tolerance for ambiguous states or unfamiliar interactions without payoff. Trust beats novelty.
Principle 3
Clerks and executives stay fully supported: distinct experiences, unified architecture.
Principle 4
Borrow patterns clerks have seen elsewhere (tables, forms, dashboard tiles), so getting productive doesn’t mean relearning railway-only conventions from scratch.
Information architecture
Before screen design, a site map made the separation explicit: clerks moving between individual orders, supervisors scanning aggregates, same product.
Site map. Primary navigation branching into clerk workflows and supervisor reporting without forcing one mode onto the other.
Concept wireframes
Four screens covered both user types before visual design: operational dashboard, order queue, shipment form, and reporting.
Today’s Dashboard
Order Queue
Shipment Entry Form
Reports View
The platform could not pull from the additional 12+ research databases clerks used daily. Improvements had to work inside real data limits.
Design for meaningful gains with the data we could reach, not a fantasy of total integration.
These were career operators: unionized clerks in the green screen for full shifts, many for years. New hires still had to ramp fast, and everyone would get trained, but the point was to cut that training way down from what the terminal demanded.
Keep the depth power users lean on; lose the months-long learning curve. Familiar SaaS patterns over railway-only convention.
High fidelity kept the data richness clerks need, but put hierarchy first. A dark NS header anchored navigation. Teal visualization made charts readable. White space, a deliberate reversal of the terminal, gave the work room to breathe.
Dashboard
Today’s Shipments surfaces what needs action now. Priority and status read at a glance, so the first screen is a triage tool, not a blank command line.
Orders
Order management borrows SaaS list patterns clerks already know: filter, sort, open, act. Density stays high; hierarchy stays honest.
Entry
Shipment entry replaces cryptic sequences with guided fields and clear validation, including edge cases like hazmat and live refrigeration status.
Reports
Customer and operational reports give supervisors trends and exceptions without assembling the picture by hand after the shift.
From sketch to structure
Early ideation mapped the railyard’s ontology onto spatial layouts before visual design locked the brand language.
High-fidelity screens
The bar for modernization
‘Measurably easier.’
Outcome · Concept · handed off at design
A Fjord engagement, researched and designed end to end, then handed to Norfolk Southern to build.
Metrics the design was built to move:
Shipping clerks
Ramp time
The terminal took months to learn. Built on patterns clerks already know, the bet is a new hire productive in days, the first number Norfolk Southern can watch.
Order accuracy
Error rate
Guided entry and inline validation exist to catch what the green screen let through. Track corrected-after-submit rate against the old baseline.
Supervisors & executives
Time to signal
Live dashboards replace reports built by hand. Measure how fast leadership sees an exception now versus the old spreadsheet lag.
The reframe: for people who’ve lived in a system for years, ‘modern’ was never the goal; ‘measurably easier than what they’ve got’ was. Every screen got judged against the green screen.