amer.catic@projectvisit.org dagb@projectvisit.org +46 31 322 7207

Designing a Kanban board for agile hardware development

Hardware development is not a linear march from idea to product. Engineering teams must shape options, test prototypes, wait for bought-in components, satisfy certification rules and learn what customers will accept. A Kanban board can make that tangled work visible, but only if it reflects actual constraints rather than presenting a tidy office workflow. For agile hardware, the board should coordinate decisions, evidence and flow across disciplines.

The design should combine disciplined pull, work-in-progress limits and clear card policies with the flexibility needed for experiments. It should also connect visual management to PLM, ERP, MES and supplier information without becoming another disconnected task tracker. The result is more than a status screen: it is a shared operating model for learning and delivery.

Why hardware needs a different kind of board

In software, a completed feature often marks a finished outcome. A physical product may still need tooling, calibration, regulatory evidence, supplier confirmation and field validation. The board therefore needs to distinguish a commitment from activity. A card should represent an outcome or decision that can move through the product-development system, while uncertainty, blocked dependencies and experimental learning remain visible.

Research by Celean and the Chalmers Electronic Lean Research Team offers useful evidence for this approach. The collection of research publications explores digital visual planning, Kanban, visual project management and enterprise integration, including industry collaborations associated with Volvo. The relevance for hardware is the emphasis on live information: a card earns its place when it supports a decision, reveals a bottleneck or records evidence needed by the next team.

The board should not reward the appearance of constant motion. Fast movement of an ambiguous card can hide rework. Useful visibility comes from seeing where work waits, which decisions remain open and whether the current flow matches product and programme priorities. This shifts attention from individual busyness to sustainable flow.

Shaping cards around real work

A strong card is a compact contract rather than a container for every file. Useful fields include the customer or business outcome, accountable owner, current maturity, acceptance evidence, linked configuration records, supplier dependencies, target date, risk and blocked time. Large teams need card templates for discovery, design, procurement, build, verification and validation. Each template can ask different questions while retaining a common identity.

Swimlanes should describe stages in the value flow, not an organisation chart. Concept, simulation, detailed design, long-lead procurement, assembly, verification and release are often more useful than “engineering” and “operations”. Cards should advance only when agreed exit criteria are met. This exposes unresolved work without forcing every technical task onto the board or pretending that design and production are separate projects.

Use a hybrid view. A large physical board helps a Melbourne hardware team gather around the same artefacts, inspect constraints and make trade-offs; a digital view supports remote suppliers, photographs, models, test results and filtered views by variant. The physical display should prompt conversation and decisions, while the digital record remains the source for detailed evidence.

Engineering flow with pull and limits

Work-in-progress limits are the mechanism that turns a board into a pull system. Set limits for each stage and, where useful, for each team or subsystem. A limit should reflect actual attention and downstream capacity, not an arbitrary target or a headcount calculation. When a lane reaches its limit, the team completes, reworks or removes work before accepting more. The visible queue then exposes capacity problems early.

Different classes of work need different rules. A safety or regulatory task may require expedited treatment, a customer variation may follow a standard flow, and a research spike should not consume production capacity. Expedite lanes need explicit entry and exit conditions so urgency does not create a shadow queue. Blocked cards should carry an owner and a next action, not simply a red label that disappears from the main view.

Flow also depends on replenishment decisions. Team members compare demand with capacity and pull the most valuable ready item, while a replenishment review handles large commitments and exceptions. Short, regular conversations work better than heroic handovers. Australian teams may say “let’s park that” when a topic needs separate attention, but parking it on the board should still preserve ownership, context and a planned return.

Connecting the board to enterprise systems

Integration should reduce duplicated entry, not create a second version of the product record. A card can reference an item in PLM, an order in ERP, a build record in MES and a software task in an engineering backlog. Events should update the card when evidence changes. Conversely, opening the card should take the engineer to the authoritative record, with permissions and configuration data handled by the source system.

The integration model should distinguish status from truth. Moving a card to “tested” because someone changed a dropdown is weak governance. It becomes stronger when the change follows an accepted test result, approved drawing revision or supplier confirmation. Link the relevant evidence and event, then let the board show it. This approach supports digital visual planning while preserving auditability for safety-critical work.

The coordination metaphor associated with Ampersand & Dance is helpful for this purpose: work has a rhythm, a sequence and shared dependencies. A board that reveals only percentages breaks that rhythm; one that shows commitments, readiness, dependencies and flow lets the team adjust timing together. The software matters less than the quality of the shared attention it enables.

Adapting the method for Australian teams

Australian hardware businesses often operate across several realities at once. Melbourne has a dense advanced-manufacturing and start-up ecosystem, yet a small product team can be spread across suburban sites, universities and contract manufacturers. Brisbane suppliers in aerospace and defence may work under formal quality systems and security restrictions. These conditions favour clear ownership and auditable links rather than an informal whiteboard that only makes sense in one room.

Supply chains add another local layer. Australia’s reliance on imported components, long shipping times and smaller local supplier base can make procurement risk central to delivery. Teams in Perth serving mining and energy customers may also work across remote sites, where connectivity and field feedback are uneven. A board should therefore expose supplier lead time, site constraints and validation status, while giving people a simple way to flag uncertainty.

The background of Celean’s industry-oriented research is summarised through Project Visit background. That experience is useful to Australian organisations, but it should be translated rather than copied. Manufacturing mix, regulatory expectations, working hours and collaboration culture differ, so pilot teams should adapt card definitions and policies to the product, the supplier network and the people who will actually use the board.

Practical recommendations for implementation

A sound design is easiest to test as a behaviour change rather than a software purchase. The following recommendations provide a practical sequence for teams beginning with limited resources.

After a short pilot, compare the new flow with the baseline and assess whether teams can identify constraints, release work more predictably and preserve required evidence. Refine the board from observed behaviour. Removing low-value fields, clarifying ambiguous states and adjusting limits often produce more benefit than adding another colour, chart or notification.

Keeping the board useful over time

Ownership must sit with the team using the board, while enterprise architects guard shared data and system boundaries. A named owner should review card templates, policies and integrations on a regular cadence. New starters need to learn not only where to click, but why work enters a lane, what evidence allows it to leave and how exceptions are handled. Training in pull, flow and service classes is as important as interface familiarisation.

The measures on the board should remain balanced. Throughput and lead time show flow, while defects, test pass rates and customer outcomes show whether speed is useful. If teams push work faster but quality falls, the board is producing a false signal. Periodically compare local performance with programme outcomes, then allow the team to change policies when the evidence supports it.

The enduring lesson is that the best Kanban board is not the most elaborate display. It is the shared, evidence-based control mechanism that helps a hardware team move safely, learn quickly and deliver what customers actually need.