Show me how

Designing a role-based workflow automation platform to replace manual, PDF-driven task management across a distributed retail operation.

Client

Bunnings Retail

Services

Discovery Workshops Workflow Mapping Prototype Testing

My Role

Lead Product Designer

Date

October 2025

01 - The Problem

A broken process masquerading as a content problem

Bunnings - one of Australia's largest retail chains - managed task completion, allocation, and compliance data collection across hundreds of stores through shared PDFs and manual handoffs. The result was a system riddled with process gaps: tasks fell through the cracks, managers had zero visibility into completion status, and there was no structured feedback loop to improve operations over time.

On the surface this looks like a software problem. In reality, it was a broken workflow — and the solution required designing a genuine automation layer on top of how a distributed workforce actually operates.

3 Distinct role types with fundamentally different workflow needs

600+

Research insights synthesised into actionable design themes

2 Rounds of usability testing before development handoff

0 Paper-based PDF workflows remaining after launch

02 - Discovery

Mapping the broken workflow before touching any UI

We began with a structured discovery phase: stakeholder workshops, contextual interviews with floor staff, store managers, and regional leaders, and detailed workflow mapping sessions. The goal wasn't just to understand pain points - it was to map the full lifecycle of a task from creation to completion, and identify every handoff, gap, and assumption baked into the current process.

The questions that shaped the discovery

Where does information break down in the current process - and at which handoff point?

What does "task completion" actually mean to each role? The answer was different for all three.

What data matters for oversight vs execution? Conflating these was a root cause of the PDF chaos.

What metrics do senior stakeholders need to act on? Error rate and TTC surfaced as the primary targets.

What makes a floor worker trust a digital instruction the same way they trust a manager speaking to them?

We synthesised 600+ insights into actionable themes, identifying three distinct mental models operating within the same workflow chain. This became the foundation for everything that followed.

Key insight: The failure mode wasn't that staff were skipping tasks. It was that the system gave them no clear signal of what to do next, no confirmation that what they'd done was received, and no way to flag blockers. The workflow had no feedback loop. Designing that loop was the central design challenge.

03 - SYSTEMS THINKING

Role architecture before screen design

One of the most consequential design decisions in this project happened before a single wireframe was drawn: establishing a clear role architecture. Each user group occupies a different position in the workflow chain, with fundamentally different interaction models, information needs, and success criteria.

Floor Staff

"What do I do next, and how do I do it right?" Simple, mobile-first execution. Step-by-step guidance. Zero ambiguity. No cognitive overhead.

Store Manager

"Is my team on track? What needs attention now?" Task assignment. Team status at a glance. Deadline visibility. Quick escalation paths.

Regional Manager

"Are my stores performing? Where are the risks?" Cross-store dashboards. Compliance trends. Error rates and time-to-completion data. Actionable signal, not raw noise.

04 - WORKFLOW DESIGN

Designing the automation loop end-to-end

The core design challenge was mapping and then encoding the full task lifecycle as an automated workflow — replacing every manual handoff with a designed interaction. Each stage of the loop needed to feel effortless for the person at that node, while reliably passing structured data to the next.

04 - WORKFLOW DESIGN

Designing the automation loop end-to-end

01

MANAGER ACTION

Task Created & Configured

Manager creates a task using a structured form. Smart defaults pre-populate common fields based on role, store, and historical task patterns, reducing repetitive input.

02

SYSTEM LOGIC

Role-Based Routing & Assignment

The system routes the task to the appropriate staff member based on role, availability, and store context. No manual distribution required.

03

TRIGGER NOTIFICATION

Staff Receives Assignment

Floor staff receive a clear, actionable notification with deadline, priority, and all context needed to act immediately without clarification.

04

FLOOR STAFF ACTION

Step-by-Step Execution

Mobile-first interface guides staff through each step of the task. Designed for speed, minimal reading, and clarity in a busy retail environment.

05

DATA CAPTURE

Completion with Structured Data

On completion, the system captures structured data (time, outcome, errors flagged) forming the data layer that powers management visibility.

06

MANGER VIEW UPDATED

Real-Time Status Visibility

Store managers see task status update in real time. No chasing, no manual check-ins. The system surfaces what needs attention proactively.

07

PROCESS INTELLIGENCE

Regional Analytics Aggregated

Completion data feeds into cross-store dashboards, giving regional managers the process intelligence to track compliance trends, error rates, and performance over time.

We used lo-fidelity wireframe workshops to map these flows with stakeholders before committing to any UI direction. This meant every screen we ultimately designed had a validated purpose in the workflow, not just a well-intentioned guess.

05 - TESTING & ITERATION

Two rounds to get the workflow right

We built a functional prototype and ran two structured rounds of usability testing with real Bunnings staff across roles. Each round had a clear objective: Round 1 was diagnostic, Round 2 was validating.

ROUND 1 - PROBLEMS FOUND

No feedback on submission state

Users had no way to tell if their task submission had processed. Several attempted to resubmit, creating duplicate records and eroding trust.

ROUND 2 - RESOLVED

Loading indicators + confirmation states

Added visible loading indicators at every async action. Confirmation screens clearly communicated successful processing, eliminating resubmission behaviour.

ROUND 1 - PROBLEMS FOUND

Ambiguous action labels causing hesitation

"Submit" created decision anxiety. Users weren't sure if it was saving a draft or completing the task. Several abandoned mid-flow.

ROUND 2 - RESOLVED

Action-specific, consequence-clear labels

Replaced generic labels with context-specific language: "Mark as Complete", "Save for Later", "Assign to Team." Abandonment on this step dropped to near zero.

ROUND 1 - PROBLEMS FOUND

Filter menus required too many taps

Managers applying filters to their task view had to navigate 3-4 taps per filter. On a busy shop floor, this was a deal-breaker for adoption.

ROUND 2 - RESOLVED

Inline filters with auto-close on selection

Redesigned filter interactions to apply immediately on selection and auto-close the panel. Reduced filter application to a single tap. Managers reported the dashboard "finally felt fast."

"Completing a request was much quicker this time. I knew exactly what to do, and I didn't have to guess what the next step was. The process feels a lot more intuitive and less frustrating."
- Bunnings Floor Staff, Usability Testing Round 2

Testing primary workflows and pinpointing elements with issues for resolve.

Showcasing and designing for how different user groups will interact with each other in a complex work flow. The above flow captures a manager assigning a task and floor staff receiving a notification and viewing the task.

After solving testing issues, we shipped!

Final Product

The second round of usability testing revealed noticeably fewer and less severe issues, confirming that the design changes had successfully addressed the main pain points. With the core workflows validated, I focused on polishing micro-interactions and improving overall ease of use.

Added visible loading indicators when submitting requests or applying filters, giving users clear feedback that their actions were being processed.
Set key form defaults (such as commonly used categories or departments) to streamline frequent tasks and reduce repetitive input.
Refined button and menu labels to provide more specific action cues, further reducing ambiguity. Updated interactions so that filter and search menus would close automatically once a selection was made, creating a smoother, more seamless experience.

“Completing a request was much quicker this time, I knew exactly what to do, and I didn’t have to guess what the next step was. The process feels a lot more intuitive and less frustrating.”

These final refinements helped ensure the portal not only addressed previous pain points but also delivered a user experience that felt easy, efficient, and fit for purpose.