B2.1.12 Iterative Development

What is modelling in iterative development?

Modelling is the process of creating a physical or digital representation of a design idea that can be tested. In the context of iterative development, a model is not a finished product — it is a tool for learning.

The most important principle of iterative modelling is:

Early models should be made quickly and cheaply — prioritising the ability to test a specific design question over visual perfection or material accuracy. The goal is to learn as quickly as possible, not to impress.

Types of Models in Iterative Development

Different types of models serve different purposes at different stages of the iterative development cycle:

Appearance Models (Aesthetic Models)

Purpose: To explore and communicate the visual and tactile qualities of a design — its form, proportion, colour, texture, and aesthetic character.

Characteristics:

• Made from materials that are easy to shape and finish quickly — foam, cardboard, clay, 3D-printed polymers

• Not necessarily functional — do not need to work mechanically

• Scaled accurately to represent the intended product dimensions

What they test:

• Does the form feel visually appropriate for the context and user?

• Does the proportion feel right in the hand or environment?

• Does the aesthetic communicate the right qualities — domestic, clinical, professional, friendly?

• Does the size feel correct?

Real-World Example:

When OXO was developing the form of the Good Grips handle, their design team carved multiple handle profiles from dense foam — quickly testing different diameters, cross-sections, and surface textures by placing them directly in the hands of research participants. These foam appearance models could be made in minutes, enabling dozens of form variations to be tested in a single session.

Functional Models (Working Models)

Purpose: To test the mechanical performance and operational function of a design — how it moves, how forces are transmitted, how mechanisms operate.

Characteristics:

• Made from materials that can simulate the mechanical behaviour of the intended design

• May use substitute materials if final materials are expensive or difficult to work with

• Focus on mechanism, structure, and function rather than surface finish

What they test:

• Does the mechanism operate as intended?

• Does the structure withstand required forces without failure?

• Does the design function correctly through its full operational range?

• What forces are required for operation?

Real-World Example:

Microsoft's development of the Xbox Adaptive Controller required extensive functional modelling of the external port connection system — the mechanism through which users plug in diverse external switches and controllers. Working models were constructed using rapid-prototyped components that replicated the mechanical connection geometry, allowing engineering teams to test insertion and removal forces, connector durability, and one-handed connection operation across many iterations before committing to final manufacturing tooling.

Ergonomic Models (User Interaction Models)

Purpose: To test the interaction between the design and the human body — how the design fits, how it is held and operated, what physical demands it places on the user, and how comfortable it is in use.

Characteristics:

• Scaled accurately to human body dimensions

• Made from materials that simulate the tactile properties of the intended materials

• Designed to be tested directly by representative users rather than evaluated by the designer alone

What they test:

• Does the design fit within the range of user body dimensions identified in the specification?

• Can the design be operated with the force levels specified?

• Does extended use cause discomfort, fatigue, or pain?

• Can the design be operated by users with the physical characteristics of the persona?

Real-World Example:

When Humanscale was developing their Freedom Chair ergonomic office seating, their development team created ergonomic models at multiple stages — each representing a specific seat geometry iteration — that were tested by participants across the full range of body sizes targeted by the specification. Each model was instrumented with pressure mapping sensors that recorded the distribution of body weight across the seat surface — generating quantitative data about ergonomic performance that annotation-based evaluation alone could not provide.

Concept Models (Low-Fidelity Prototypes)

Purpose: To quickly explore and communicate a design concept at an early stage — before significant development investment has been made. These are sometimes called "rough prototypes" or "sketch models".

Characteristics:

• Made very quickly from whatever materials are most accessible — cardboard, tape, foam, paper, clay

• Often crude in appearance — representing the idea of the design rather than its final form

• Intentionally impermanent — expected to be modified, rebuilt, or discarded

What they test:

• Does this concept make sense when made physical?

• Is the basic idea viable — does it fundamentally work?

• Which of several competing concepts shows the most promise?

Real-World Example:

When IDEO was developing a redesigned shopping trolley for a major supermarket, their first concept models were made entirely from existing trolley components, cable ties, and cardboard — assembled in a single afternoon. These crude concept models allowed the team to test fundamental ideas about compartment organisation, child seating integration, and wheel mechanism before investing any resources in refined development.

One concept model — a trolley with a wire basket that detached from the frame to serve as a carry basket — was made from repurposed components and tested in an actual supermarket within 24 hours of the initial brainstorming session. The test revealed immediately that the concept was unworkable in practice — a learning that would have taken weeks to discover through drawing-based development alone.

Digital Models (CAD and Virtual Prototypes)

Purpose: To create precise, dimensionally accurate representations of a design that can be analysed, simulated, and communicated digitally — and in many cases, directly manufactured through digital fabrication technologies.

Characteristics:

• Created using Computer-Aided Design (CAD) software

• Can be analysed for structural performance, material properties, and manufacturing feasibility

• Can be rapidly translated into physical models through 3D printing, laser cutting, or CNC machining

• Enable precise dimensional checking against specification criteria

What they test:

• Are the dimensions precisely correct and consistent?

• Does the design meet structural performance requirements under simulated loading?

• Can the design be manufactured using the intended processes and materials?

• Does the design assemble correctly when all components are considered together?

Real-World Example:

When Dyson develops new universal design products — such as their lightweight hair dryers designed for users with limited arm strength and mobility — their digital modelling process uses finite element analysis (FEA) to simulate how the product's internal structure responds to the forces applied during use. This digital testing allows structural performance to be evaluated and refined across hundreds of design iterations in the time it would take to physically build and test a single prototype.

Digital models are then used to generate 3D-printed physical prototypes for ergonomic and user testing — combining the speed and precision of digital development with the essential realism of physical user interaction.

What does testing mean in iterative development?

Testing is the systematic process of evaluating a model against specific, defined criteria — generating evidence about what the design does well, what it fails to do, and exactly where and why it falls short.

Effective testing in iterative development has three essential characteristics:

Characteristic 1: Testing Is Purposeful

Every test should be designed to answer a specific question — derived from the current stage of development and the gaps identified in the previous iteration.

Purposeful testing generates actionable data. Unfocused testing generates general impressions.

Characteristic 2: Testing Involves Real Users

The most valuable tests involve representative users — people who match the characteristics of the persona — interacting with the model under realistic conditions.

Designer self-evaluation — testing the model yourself — has significant limitations:

• Designers do not have the physical characteristics of the target user

• Designers know how the design is intended to work, which unconsciously compensates for interaction difficulties

• Designers have emotional investment in the design that can bias their evaluation

Real user testing removes these biases and reveals how the design actually performs for the people it is designed for.

Characteristic 3: Testing Generates Both Qualitative and Quantitative Data

Effective testing captures both types of data:

Quantitative data tells you whether the design meets the specification. Qualitative data tells you why it does or does not meet the user's needs.

Testing Methods in Iterative Development

Recording Test Results

Test results must be systematically recorded to be useful for the refinement phase.

Effective test recording includes:

• Quantitative measurements presented in tables or charts

• Qualitative observations presented as annotated notes or transcripts

• Photographic or video documentation of user testing sessions

• Direct comparison of test results against relevant specification criteria

• Clear identification of what passed, what failed, and what was inconclusive

What does refinement mean in iterative development?

Refinement is the process of using test findings to make specific, targeted improvements to the design — modifying what failed, enhancing what partially succeeded, and preserving what worked well.

Refinement is not redesign from scratch. It is focused, evidence-based improvement — changing the minimum necessary to address identified failures while preserving the elements that are already working.

Principles of Effective Refinement

Principle 1: Address Critical Failures First

Refinement resources — time, materials, fabrication capacity — are always limited. Prioritise changes that address essential specification failures and critical user need gaps before addressing desirable criteria or enhancement opportunities.

Principle 2: Change One Variable at a Time

When multiple refinements are made simultaneously, it becomes impossible to determine which change caused which improvement in subsequent testing. Where possible, make single-variable changes between iterations — changing one aspect of the design and testing again before making additional changes.

This scientific approach to refinement generates much clearer causal understanding of the relationship between design decisions and performance outcomes.

Principle 3 : Document Every Refinement Decision

Every change made during refinement must be recorded with its justification — connecting the change explicitly to the test finding that prompted it.

This level of documentation creates a traceable design history — an evidential record of how the design evolved in response to testing, demonstrating genuine iterative thinking.

Principle 4 : Know When to Refine and When to Rethink

Not all refinement involves small adjustments. Sometimes testing reveals that a fundamental aspect of the design concept is flawed — that the core approach cannot meet the specification requirements regardless of refinement.

In these cases, iterative development requires the courage to return to ideation — to go back to the concepts generated in B2.1.10 and explore a different design direction. This is not failure — it is the design process working exactly as it should.

Dyson's development of progressively lighter vacuum cleaners — driven by research showing that older users and users with upper limb weakness struggled with traditional heavy models — involved extensive iterative development cycles:

• Round 1 models: Full-scale foam appearance models testing weight distribution and handle geometry

• Round 2 models: Functional models testing motor-to-filter airflow geometry with working motors

• Round 3 models: Ergonomic prototypes tested with elderly participants — measuring arm fatigue over timed cleaning tasks

• Round 4 models: Pre-production prototypes tested for durability — 500 use cycles across diverse floor surfaces

Each round of testing generated specific findings that drove targeted refinements — reducing weight from initial 4.2kg through successive iterations to a final production weight of 2.68kg, while maintaining suction performance and meeting all safety and durability specification criteria.

The development of Caring Cutlery — ergonomically designed cutlery for people with Parkinson's disease, arthritis, and tremor conditions — involved a model-test-refine cycle that revealed unexpected insights at each iteration:

Iteration 1 test finding: Heavier handles — initially specced to counteract tremor — actually increased user fatigue and wrist pain over extended meal times. Specification was revised.

Iteration 2 test finding: Increased handle diameter — developed to improve grip — caused difficulty fitting cutlery in standard dishwasher baskets. Secondary specification criterion added.

Iteration 3 test finding: Soft rubber grip surface — developed for tactile security — absorbed food odours after repeated use. Material substituted for textured hard polymer.

Iteration 4 test finding: Final ergonomic profile tested positively with Parkinson's users but was found to require an adapted grip position incompatible with how stroke survivors held cutlery. Two handle variants developed.

Each unexpected finding — none of which could have been predicted from the initial specification — was revealed only through physical modelling and real user testing. The model-test-refine cycle was not just how the design improved — it was how the design learned what it needed to be.