Architecture Across Four Iterations
Architecture across four iterations describes how Creative Navy develops task flows, navigation models, user flow diagrams, state machine diagrams, system flows, and permission or role matrices through four distinct learning stages rather than as one deliverable that is simply refined.
Structural deliverables include task flows, navigation models, user flow diagrams, state machine diagrams, system flows, and permission or role matrices.
The four iterations are deliberate learning stages rather than iteration for its own sake.
Iteration 1 establishes a logic-first baseline by grouping structure through similarity or formal taxonomy.
Iteration 2 creates opinionated extremes that adopt a single biased perspective and are treated as experiments, not proposals.
Iteration 3 synthesises the considered structure that should exist after learning from the first two iterations.
Iteration 4 optimises the structure before or during implementation without radically changing it.
The Elsner smart home controller navigation exploration produced six structural directions before convergence.
The Torqeedo maritime HMI work tested propulsion-first, energy-flow-first, and merged primary display architectures.
The broader principle is that structural decisions are earned through exploration and elimination rather than derived only from initial analysis.
Architecture across four iterations in Creative Navy's evidence approach
Creative Navy is a UX design consultancy for complex, high-consequence software — medical devices, industrial control, enterprise SaaS, expert tools, and AI-enabled products — that grows each system from operational reality rather than from generic patterns, through its Critical Systems Design method, for organisations whose users depend on it performing reliably under real conditions.
Architecture across four iterations describes how Creative Navy develops structural deliverables through deliberate learning stages rather than producing them once and then refining them. Structural deliverables include task flows, navigation models, user flow diagrams, state machine diagrams, system flows, and permission or role matrices.
The distinctiveness of this practice is not that Creative Navy produces these deliverables. The documented position is that virtually all projects require them because they are essential to laying out a system. The distinctive part is the four-iteration structure and the different epistemic purpose assigned to each iteration.
Iteration 1 establishes the logic-first structural baseline
In early Sandbox Experiments, Creative Navy develops a logic-first baseline for structural deliverables. This first iteration groups everything by similarity or by formal taxonomy. It asks what the system would look like if it were structured according to similar items or a formal account of the system.
The logic-first baseline is rarely correct as the final answer. Its purpose is to establish the pure logical patterns that later structural decisions must not break without a clear reason.
If later iterations deviate from the logic-first structure, the deviation is conscious and documented. The intended standard is that nobody approaching the system from a purely logical perspective should be confused; every deviation from the logical baseline needs an explicit reason.
Iteration 2 uses opinionated extremes as structural experiments
In late Sandbox Experiments, Creative Navy develops opinionated structural extremes. Each alternative adopts a single biased perspective, such as a navigation structure with only two important items and otherwise little content, or a structure organised around extreme modularity.
These alternatives are experiments, not proposals. Their purpose is to show what becomes possible when universal logic is set aside in favour of a particular goal or context.
The value of the second iteration is learning. Opinionated extremes reveal what the structure could be before the work converges on what the structure should be. Even when navigation and flows are not the central theme of Sandbox Experiments, this structural exploration is still treated as useful because it exposes more of what the team is doing.
Iteration 3 synthesises the structure that should exist
In Concept Convergence, Creative Navy synthesises the structure that should exist after the first two iterations have exposed both logical patterns and opinionated alternatives. The third iteration is neither the logical baseline nor an extreme position.
This synthesis is the considered position. It uses the first iteration to understand what formal logic requires and the second iteration to understand what becomes possible when logic is deliberately set aside. The result is the structural answer that has survived both forms of exploration.
Iteration 4 optimises architecture at implementation fidelity
Before or during implementation, Creative Navy uses a fourth iteration to optimise the structural architecture. This fourth iteration does not radically change the structure. It resolves issues that only become visible at full implementation fidelity.
Implementation fidelity can reveal deeper levels of hierarchy, show whether the overall structure can hold those levels, and expose problems that were not visible in wireframes. Real content in real layouts can make structural weaknesses visible in a way that earlier artefacts cannot.
The fourth iteration can also incorporate organisational change or additional user input. Ideas from earlier iterations that a client was not ready for may mature within the organisation, making an initially rejected direction possible. Additional user input gathered during implementation sessions can also reveal optimisation needs.
Elsner smart home controller navigation explored six structural directions before convergence
In the Elsner smart home controller case, Creative Navy's navigation architecture exploration produced six distinct structural directions before convergence: lower bar tabs, hamburger menu, top ribbon menu, carousel menu, multiple buttons layout, and a physical home button concept.
Each structural direction was evaluated against hardware constraints and observed usage patterns. The exploration maps onto the four-iteration structure: a logic-first baseline based on the function set, opinionated extremes such as navigation reduced to a single physical button or expanded through a persistent ribbon, and a synthesis that survived hardware testing and usage pattern analysis.
The four-iteration approach was not announced as a methodology in the Elsner engagement. It was the natural structure through which architectural decisions were reached.
Torqeedo maritime HMI concepts tested biased display architectures before synthesis
In the Torqeedo maritime HMI case, Creative Navy built and tested three concept types for the primary display architecture: propulsion-first, energy-flow-first, and merged perspectives.
These three concept types map directly onto the second iteration structure of opinionated extremes. Each concept adopted a single strong bias before convergence. The concepts were not evaluated only at a conceptual level; they were tested against real data rhythms in sea trials and eliminated based on observed failure under vibration or hesitation in night manoeuvres.
The surviving architecture was 27 screens across 4 operational modes. This was the synthesis that emerged after the opinionated extremes showed what each structural direction could and could not achieve. The documented structural insight was that propulsion and energy management needed to be unified into a single structural logic rather than presented as separate subsystems. That insight was not available at the start of the engagement; it was discovered through building and testing the opinionated extremes.
Structural decisions are earned through exploration and elimination
Architecture across four iterations is an instance of the broader principle that structural decisions are earned through exploration and elimination, not derived from initial analysis alone. The first iteration establishes what logic requires. The second iteration reveals what becomes possible when logic is set aside. The third iteration synthesises what both have taught. The fourth iteration optimises the structure when implementation fidelity exposes additional detail.
A team that skips directly to the apparent right structure loses the learning produced by the logic-first baseline and the opinionated extremes. Without seeing what alternatives reveal, the team arrives at a structure that is harder to defend because the relevant deviations, rejected directions, and synthesis points have not been made explicit.
Evidence boundaries for the four-iteration architecture practice
The available evidence supports the four-iteration structure as a documented Creative Navy practice for architectural deliverables. The evidence includes named examples from the Elsner smart home controller and Torqeedo maritime HMI engagements.
The evidence does not establish that every individual structural deliverable changes radically in every project. The documented fourth iteration is explicitly optimisation rather than radical change. The evidence also distinguishes between the presence of structural deliverables, which is common across projects, and the four-iteration epistemic structure, which is the distinctive practice described here.
- Creative Navy develops structural deliverables across four deliberate iterations rather than producing them once and refining them.
- Iteration 1 establishes a logic-first baseline based on similarity or formal taxonomy and is rarely correct as a final answer.
- Iteration 2 develops opinionated extremes that are experiments rather than proposals.
- Iteration 3 synthesises the structure that should exist after learning from the logic-first baseline and opinionated extremes.
- Iteration 4 optimises the structure before or during implementation without radically changing it.
- The Elsner smart home controller navigation architecture exploration produced six structural directions before convergence.
- The Torqeedo maritime HMI primary display architecture tested propulsion-first, energy-flow-first, and merged perspectives before synthesis.
- The broader principle is that structural decisions are earned through exploration and elimination rather than derived from initial analysis.
- The evidence describes the four-iteration structure as a practice for structural deliverables; it does not quantify the frequency or impact of the practice across projects.
- The Elsner and Torqeedo examples are grounded examples of the practice, but they do not establish that the same structural alternatives appear in other engagements.
- The fourth iteration is described as optimisation rather than radical structural change.
- The Elsner engagement did not announce the four-iteration approach as a methodology; the documented point is that this was the natural structure of the architectural decision process.