Discoverability rides on prior knowledge and signifiers
Aliases: affordance · product semantics
What it is
Screenless devices have exactly two sources of feature discoverability: prior knowledge — expectations users carry from past products and conventions (clockwise increases, double-tap toggles, press-and-hold pairs); and signifiers — deliberately placed perceivable traces on the device or in the environment that an operation exists (graphics, shape, texture, labels, light and sound feedback).
There is no third route. The on-screen list is one high-density species of signifier; screenless devices, having lost it, are left with two low-bandwidth sources — so the discoverability work of screenless design is essentially how well the design parasitises prior knowledge and which correct signifiers it leaves on the device.
Keep "affordance" and "signifier" distinct: an affordance is the action possibility that actually exists between object and actor (the lamp can be voice-controlled — an affordance); a signifier is the perceivable signal that communicates the possibility. Screenless devices never lack affordances — they lack signifiers. The capability exists; the trace does not.
Why it happens
How each source does its work:
- Prior knowledge is a default expectation package. Given any new object, users first retrieve operation conventions from memory of similar objects: switches press, knobs turn, sliders slide. These conventions come from decades of physical use and are strong and automatic. The design lever is grafting new capabilities onto old conventions — "twist the lamp base to dim" needs almost no learning, because the twist-dims mapping is pre-installed. Conversely, designs that violate convention (a knob where a press means latch-on) do not merely cost learning time — they create a standing wrong expectation the user re-enacts on every encounter.
- Signifiers are externalised memory. Form (the concavity of a button), texture (grip ridges), marking (a voice-entry icon), sound (the click of an effective press) all say "something can be done here". The bandwidth is tiny — a device carries at most a handful of independent signifiers — so they must be reserved for a critical few operations, and must point the same way as prior knowledge (icons matching universal conventions), or the signifier itself needs interpreting.
Together they set the discovery rate: fit with prior knowledge × presence of signifiers = the probability a first-time user finds the operation. Either factor at zero drives discovery to zero — the strongest capability remains dark matter.
Studying it
- Norman's affordance/signifier framework (The Design of Everyday Things) is the theoretical source of the distinction, with systematic treatment of usability failures where an affordance exists but no signifier communicates it.
- First-use observation: hand the new device, without manual, to a user who has never seen it, and record the first five minutes of spontaneous actions — what was attempted, in what order, when help was sought. This directly measures convention fit (zero correct actions in five minutes = zero fit).
- Convention surveys: sample the target population's holding rates of common interaction conventions ("what does a double-tap mean", "what does press-and-hold do") to inform parasitism choices — convention packages differ markedly across generations and cultures and cannot be assumed.
One methodological caution: the lab first-use five minutes are a high-motivation state; real households bring lower motivation and less exploration. Field discovery rates are systematically lower than observed — a standing overestimation of this method.
Where it stops holding
- Prior knowledge cuts both ways. Parasitising old conventions buys zero learning but inherits their ceiling — expectations of "a knob" stop at dimming, and functions grafted above it (twist + hold to enter scene editing) are discovered far less than the primary one; the stronger the convention, the more invisible whatever exceeds it.
- Signifiers wear out. Labels peel, icons no longer match new firmware, light and sound cues drown in noisy settings; signifiers need a maintenance budget and are not designed once and done.
- Generations hold different conventions. "Swipe" and "long-press" conventions have low holding rates among older adults — the same signifier set can differ by an order of magnitude in discoverability across groups, so single designs cannot cover all; layer by target population.
Applying it
- Prefer grafting new capabilities onto strong conventions: rotation for levels, press for on/off, double-tap for toggling — and always in the conventional direction (increase = clockwise / up / right); there is no reason to do otherwise.
- Spend the signifier budget on two places: the entry point of the primary operation (must be found) and safety-relevant operations (must be findable under stress); decorative signifiers get cut — a low-bandwidth channel carries no passengers.
- Prepare environment-level signifiers for screenless devices: the quick-command card on the wall, the legend sticker near the device — homely, but the correct answer under bandwidth constraints; make them swappable (a new card after a firmware upgrade).
- How to check: the five-minute new-user test — no manual, record first-attempt success rate and time before help-seeking; run it per population stratum (age, technical familiarity) before release. Any stratum with "zero discoveries in five minutes" goes back to design, not to better copy.