Interactivity needs an explicit signal

What it is

“You can use this” almost never grows out of a pane of glass. It has to be rewritten by an interactivity cue—evidence a person can see: the picture moves when a body nears; a silhouette is mirrored; a control jumps out of the content as something that can be hit. The question it answers is “can I enter now,” not “what precise first hit counts as success,” and not “did motion capture attention.” Motion can be an advertisement. A cue has to attribute the motion to you.

Why it happens

People infer from perceived affordance, fast. A public screen lacks the mouse pointer of a desktop and the icon grid of a phone, so the usual anchors are gone. What remains is anomaly: content answering, in time, to someone’s position, outline, or step. That coupling is the strongest evidence because “the programme just happened to play” cannot explain it. A written “please touch” is weaker—ads say please too, and the line has to be read first.

Cues also have a direction. A mirrored outline, a floor sensing zone, a large target within arm’s reach, all point at an input the body can already perform. If the cue demonstrates a fine mid-air gesture while the person stands three metres out, evidence and executable action disagree, and the classification does not flip. The cue has to land inside what this body can do at this distance, or it is only another unread animation.

Studying it

Looking Glass threw passers-by’s silhouettes back onto the screen as visual evidence of interactivity, against a no-silhouette baseline. Kindred manipulations insert an immediate reaction to a person into the attract loop, overlay an obvious touch target, or make the input site physical (a tray, a mat).

Independent variables: mirroring or position coupling present or not, verbal appeal versus bodily coupling, distance at which the cue appears. Dependent variables: first-attempt rate among those who already looked, time from seeing the cue to a reach, the share who file the cue as advertising and leave.

Lab participants already know they are facing an apparatus, so the cue’s increment flattens; the real increment is in unwitting passing-by. Cues also steal attention. Analyses must split “more people looked” from “of those who looked, more judged it usable,” or a noticing effect gets booked as discoverability.

Where it stops holding

When the input is unique and the object already looks like a machine (ATM, ticket vendor), an extra “touch here” animation is noise. A saturated media wall stages a cue contest; a single silhouette is eaten. People with limited vision may never see a mirrored outline; a visual-only cue does not create discoverability for them. If every advert in the block starts mirroring passers-by, the coupling habituates, the cue dies, and the read-only default returns.

Applying it

• Treat “answering a person” as the main cue, not “moving by itself”: state change on approach, an outline appearing, a contact highlight at once.
• Place the cue on an input available at the current distance: body outline or a step from afar, a large touch target at the glass. Do not use a fine gesture that cannot be seen from three metres and is useless once someone is at the surface.
• Keep verbal appeals secondary, and park them on the hittable spot. Do not bury “interactive experience” inside advertising copy.
• Verify by sampling people who already looked: “how did you know you could use it.” If they cannot name a concrete piece of evidence (“it followed me,” “there was an obvious button”), the cue is not working. When comparing before and after a cue, report notice rate and attempt-among-noticers separately.

Related

• Within the group: K7.02.1 People assume a public screen is not interactive · K7.02.3 The first-action threshold governs conversion
• Adjacent: K7.01 The Passing-by Effect · A5.07 Attention capture · K7.05 Near–far Interaction Switching
• Search terms: interactivity cue · attract loop · silhouette · perceived affordance