Interaction design shapes the back-and-forth between a person and a product: what the system shows, what it lets you do, and how it answers. This companion sets out what the field is, how it differs from usability and user experience, the five dimensions an interface is made of, the four activities that produce one, and the principles and goals against which a design is judged.
Interaction design is the practice of shaping the moment-to-moment exchange between a person and an interactive product: what the product makes visible, what it lets the person do, and how it responds.
The standard definition is Preece, Rogers and Sharp's: designing interactive products to support the way people communicate and interact in their everyday and working lives. The emphasis falls on interactive and on people. The subject is not the screen in isolation but the back-and-forth: a person forms an intention, acts on the product, and the product answers, which shapes the next intention. Design is good when that loop runs without friction and bad when it fights the person at every turn.
Designing interactive products to support the way people communicate and interact in their everyday and working lives.
Preece, Rogers and Sharp, Interaction Design: Beyond Human-Computer InteractionTwo things follow from the definition. First, interaction design is about behaviour, not decoration: it decides what happens when the person acts, not merely how the surface looks. Second, it is centred on the user: the measure of a design is whether real people can do what they came to do, not whether it satisfies the designer's taste. Those two commitments run through every later section of this companion.
The phrase interaction design was coined in the mid-1980s by Bill Moggridge and Bill Verplank, who needed a name for the discipline forming around the design of software behaviour rather than physical form. Moggridge had designed the first laptop computer, the GRiD Compass, and found that industrial design had no vocabulary for the part of a product you cannot touch. Gillian Crampton Smith then established the teaching of the field, first at the Royal College of Art in London and later at Interaction Design Ivrea. The discipline is the behavioural counterpart to graphic and industrial design: where those shape how a product looks and how it feels in the hand, interaction design shapes what it does in response to a person.
The international standard ISO 9241-210 frames the same work as human-centred design: an iterative process that rests on an explicit understanding of users, tasks and contexts, involves users throughout, and is driven by user-centred evaluation. The four activities in the process section below are that standard made concrete, and the evaluation chapter is its final clause.
The difference between a product that feels obvious and one that feels like a fight is rarely the feature list. It is whether the design lets you see the state of the system and matches your action to your intention.
A well-designed object tells you what it is doing and how to work it, often without a single word of instruction. A badly designed one hides its state, accepts actions that lead nowhere, and leaves you guessing whether anything happened. The same task, the same person; only the design has changed.
Durrell Bishop's 1992 answering machine is the cleanest illustration. Each incoming voice message drops a physical marble into a tray. You count the messages at a glance, because each one is an object you can see. To play a message, you place its marble in a dish. To call the person back, you set the marble on the phone. There is no screen and no menu: the object itself is the interface.
Nothing about it has to be learned. It borrows what your hands have known since childhood: the state is simply visible (you can see how many messages are waiting), and the action matches the intention (to act on a message, you handle the thing that stands for it). This is Norman's visibility and mapping, working years before Norman named them. Section iv shows this machine moving, alongside the same ideas on a phone screen.
Donald Norman models any interaction as a cycle of seven stages of action. Starting from a goal, the person plans, specifies and performs an action on the world (the execution side); then perceives, interprets and compares the result against the goal (the evaluation side). A design must support both arcs, and it fails at two specific places Norman names the gulfs.
A product feels obvious when both gulfs are narrow and feels like a fight when either is wide. The butterfly ballot in the evaluation chapter is a Gulf-of-Execution failure committed on paper; a progress bar that never moves is a Gulf-of-Evaluation failure. Bridging the two gulfs is, in one sentence, what interaction design is for.
Norman separates three things that are easily confused. The conceptual model is the designer's account of how the system works. The mental model is the user's account of how it works, assembled from experience. The two never meet directly, because the designer cannot speak to every user; everything the designer wants to convey has to travel through the system image, the visible structure of the product itself, its controls, labels, feedback and behaviour. Good design is the work of building a system image from which the intended mental model forms on its own. When the system image is poor, the user invents a mental model that is wrong, and every later action inherits the error.
A product feels obvious when both gulfs are narrow: the right action is available and visible, and the result of acting is perceivable and clear. You notice design most when it fails, because a wide gulf forces the user to think. Naming which gulf a product fails at, execution or evaluation, is the first diagnostic move of interaction design.
Interaction design sits among several overlapping terms that are often used loosely. Holding the distinctions makes the field, and what it is responsible for, much clearer.
| Term | What it covers | Relation to interaction design |
|---|---|---|
| Interaction design | The back-and-forth: what the product shows, what it lets you do, and how it answers. | The core craft. |
| User experience (UX) | The whole relationship a person has with a product or service, including moments that involve no interface at all (the packaging, the support call, the wait). | UX wraps around interaction design. |
| User interface (UI) | The concrete surface: the layout, type, colour, and controls a person sees and touches. | UI sits inside interaction design, as one of its outputs. |
| Usability | How effectively, efficiently and safely people can use the product; a measurable quality, not a discipline. | The property interaction design is trying to achieve, and the thing evaluation measures. |
A working rule of thumb: interaction design is the core craft of the back-and-forth; UX wraps around it; UI sits inside it. Interaction design shapes the moment of interaction, the five dimensions of the next section. UX is the whole relationship that surrounds that moment, including service touchpoints and feelings that involve no interface at all. Usability is then the standard against which the interaction is judged, and the subject of the evaluation chapter.
Usability has a precise standard definition. ISO 9241-11 defines it as the extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction in a specified context of use. The clause that is easy to miss is the last one: usability is not a fixed property of a product but a relation between a product, its users, their goals and their situation. The same interface can be highly usable for one group and unusable for another, which is why evaluation always fixes the users, the tasks and the context before it measures anything.
Gillian Crampton Smith named four dimensions of the language of interaction design, words, visual representations, physical objects or space, and time; Kevin Silver added a fifth, behaviour, in 2007. Together they describe the raw material an interaction is built from. The first three let a person act; the last two bring the interface to life.
Labels, button text, instructions, the terms in a menu. Words carry meaning precisely, and they are translated for every language and reading level.
Icons, images, typography, diagrams. A camera icon carries its meaning across languages; visual style (iOS rounds and outlines, Android fills and squares) sets the tone.
The device or object the person acts through, and the space it occupies: a phone, a mouse, a marble, a kiosk. The thing in the hand is part of the interface.
Media that change as you watch: animation, video, sound, and the time an action takes. A progress bar, a status that updates, a transition that shows where you came from.
Action and reaction: what the person does, and how the system answers. The rule that connects an input to its response, and the feedback that confirms it.
A useful way to hold them: words, visuals and the physical device are the nouns of an interface; time and behaviour are its verbs. The first three dimensions describe a still picture. Time and behaviour are what separate a living interface from a frozen screenshot: a button is just a shape until it responds when pressed, and a message is just text until its status changes as it travels.
Interaction design shapes the moment of interaction, these five dimensions. User experience is the whole relationship around it, including service touchpoints and feelings that involve no interface at all. When you design the five dimensions of a control, you are doing interaction design; when you account for the wait, the email that follows, and how the person felt about the brand, you are working on UX.
If the frame does not load, open it directly: interaction design, alive.
Interaction design is not a single act of inspiration but a cycle of four activities, repeated until the design holds up against real use. This is the loop ISO 9241-210 calls human-centred design.
| Activity | What it does |
|---|---|
| Discovering requirements | Find out who the users are, what they are trying to do, and the constraints, through interviews, observation, personas and scenarios. State the requirements so they can be checked later, ideally as measurable usability requirements. |
| Designing alternatives | Propose more than one way the product could work. Generating alternatives, rather than committing to the first idea, is what gives the process something to choose between. |
| Prototyping | Build a version concrete enough that a person can move through it, from a paper sketch to an interactive prototype. A prototype turns an argument about the design into something testable. |
| Evaluating | Put the prototype in front of users and measure how well it supports the task. The findings feed back: refine the design, redesign, or even rethink the requirement. |
The arrow from evaluation runs backwards: a result can send you back to the prototype, the alternatives, or all the way to the requirements. That loop is the heart of the method, and it is why the process matters more than any single rulebook. Each new form of a design breaks an assumption the last one relied on, and only evaluation tells you which assumptions held.
The four activities are not abstract; the three project phases run them almost one to one. Phase one discovers requirements (heuristic evaluation, interviews, personas, scenarios, measurable usability requirements). Phase two designs alternatives and builds the interactive prototype a user can move through. Phase three evaluates (a usability test, the System Usability Scale, a heuristic re-evaluation) and closes the loop with a revision plan.
A requirement says what must be true; a design decision chooses what the system does so that it comes true. "A first-time user can register in under two minutes without help" is a requirement. Putting the seat counter on the first screen, defaulting the term to the current one, and confirming each step are design decisions that make it come true. Read a scenario the other way and its verbs and friction moments become a design to-do list, and each item lands on a design principle.
Donald Norman's principles are the small set of ideas that recur through every good design. They are not rules to apply mechanically but lenses for seeing why a design works or fails.
| Principle | What it asks | In the marble machine |
|---|---|---|
| Visibility | Can the person see the current state and the available actions? | The waiting messages are marbles you can count at a glance. |
| Feedback | Does every action produce a clear, immediate response? | Drop a marble in the dish and the message plays; the action and its effect are one. |
| Mapping | Does the relationship between a control and its effect match expectation? | To act on a message you handle the object that stands for it, the most natural mapping there is. |
| Affordances | Does the form suggest how it is used? | A marble invites picking up and placing; nothing else needs explaining. |
| Constraints | Does the design prevent actions that make no sense? | A marble fits the dish and the phone, so the only moves are the meaningful ones. |
| Consistency | Do similar things work in similar ways? | Every message is the same kind of object, handled the same way. |
Three of these run through almost every example in this chapter: visibility (you can see the state), feedback (the system answers), and mapping (the control matches the intention). When a product feels obvious, one of these is usually being honoured; when it feels like a fight, one of them is usually being broken.
Two of these terms are routinely confused, and Norman sharpened them in the 2013 revision of his book. An affordance is a relationship between an object and a person: a possibility for action the object offers, given that person's capabilities. A chair affords sitting; a flat plate on a door affords pushing. The concept is James Gibson's, from ecological psychology (1979). A signifier, by contrast, is the perceivable cue that tells the person where and how to act: the word "push", the shape of a handle, the underline on a link. Affordances make an action possible; signifiers make it discoverable. Many failures of so-called intuitive design are objects that afford the right action but signify the wrong one, or signify nothing at all, which is why a door that must be pushed but carries a pull handle defeats almost everyone. Such doors are now called Norman doors.
Usability goals are the functional standards a design must meet. Each is best stated as a question you can put to a real user and, with care, measure.
| Goal | The question it asks |
|---|---|
| Effective to use | Can people accomplish what they came to do, and accomplish it correctly? |
| Efficient to use | Once learned, can they do it quickly and with little effort? |
| Safe to use | Does it protect people from dangerous errors, and let them recover when one happens? |
| Good utility | Does it provide the right functions for what people actually need to do? |
| Easy to learn (learnability) | Can a first-timer manage with no help and no manual? |
| Easy to remember (memorability) | After a break, can a returning user pick it up again without relearning? |
Stated this way, usability goals become measurable usability requirements: "a first-time user completes registration in under two minutes with no errors" is effectiveness, efficiency and learnability turned into a target a usability test can check. This is the bridge from interaction design to evaluation: the goals named here are precisely what the evaluation chapter sets out to measure.
Usability asks whether people can use a product. User experience goals ask how it feels to use it, the qualities of the experience rather than its mechanics.
Where usability goals are functional, user experience goals are experiential. A product can be perfectly usable and still leave a person cold, or be a small joy to use. The desirable qualities a design may aim for include being satisfying, enjoyable, engaging, pleasurable, helpful, and motivating; the qualities to design away from include being frustrating, boring, annoying, or patronising.
Satisfying, enjoyable, engaging, pleasurable, helpful, motivating, aesthetically pleasing. The qualities that make a person want to return.
Frustrating, boring, annoying, confusing, patronising. The qualities that quietly drive people away even when the product technically works.
Usability and experience are not in tension; an unusable product is rarely enjoyable. Usability is the floor, experience is what you build on top of it.
User experience goals are harder to measure than usability goals, but not immeasurable: satisfaction is captured by instruments such as the System Usability Scale, and engagement by behaviour over time. The evaluation chapter returns to exactly how these softer qualities are turned into evidence.
The works this chapter draws on, and the sources for the marble machine and the five dimensions.