Noticing-to-interaction is a multi-stage leak

What it is

People in front of a public screen are not in two bins, “saw it” and “used it.” Brignull and Rogers, and later the audience funnel, cut a continuous process into passing by, noticing, viewing, interacting. Each stage drops a cohort. Direct users are the thin remainder at the bottom, not everyone who ever registered the screen. This entry is about that layered leak: dividing interaction counts by footfall collapses failures that happened at different gates into one number. It does not re-derive why the first gate (never noticing) forms, and it does not treat “they thought it was not interactive” as the only gate.

Why it happens

Each stage filters a different thing. Passing-by to noticing is visual competition. Noticing to stopping is “is this for me.” Stopping to touching is cost, social exposure, and whether a trial is immediately possible. The stages multiply, so the bottom can be nearly empty while the actual break sits in the middle. Reading zero interactions as “bad content” or “broken hardware” repairs the wrong layer.

The funnel is also a time budget. People are walking; each stage must complete in a few strides. There is no desktop pause in which to decide later. If one stage takes too long, they have left the observation zone. Onlookers may linger at viewing without ever interacting: the count looks like attention, the conversion is still a leak. States have to be counted separately; a crowd is not a success.

Studying it

Michelis and Müller’s audience funnel and Brignull and Rogers’ passing-by / viewing / interacting codes are state-machine observation of real flow: which stage each person occupies, how long they stay, where they exit. Deployment logs need days, so the novelty bump can be split from the habituated baseline.

Independent variables: which stage is intervened (easier to notice, easier to stop, easier to start), whether the space affords lingering, whether someone is already there. Dependent variables: headcount and dwell per stage, stage-to-stage conversion, the share who leave viewing without interacting.

Do not take “interactions today” as the success metric—the bottom cannot diagnose the clog. Walking lab participants to the screen starts measurement in mid-funnel and skips passing-by and noticing. The codebook must define visible criteria for noticing and interacting (head turn, a hand reaching, a system response) or observers will credit peripheral grazes.

Where it stops holding

A reception terminal that one person is sent to (check-in, a ticket dispenser) is not a passing-by funnel: the task already named the destination. People do not leak on “whether to look”; they leak on operation difficulty. A corridor with no place to stand has no viewing stage; the funnel snaps after noticing, and copy will not rebuild a missing floor plate. An artwork may treat spectating as the outcome; scoring it by interaction rate misreads ambient success. A neighbourhood screen with return visitors has a different shape from an airport corridor; stage definitions do not transfer.

Applying it

• Log or code passing-by, noticing, stopping, and hands-on separately. Do not report only sessions at the bottom.
• When conversion collapses, locate the dying stage first. Near-zero noticing is a placement and temporal-change problem. Noticing without hands-on is a stopping-space and first-feedback problem.
• Keep spectator counts in their own column; do not merge them into “successful use.”
• Verify by drawing the funnel across several weekdays in the same clock window, comparing stage conversion rather than raw interactions. After changing one layer, expect outflow to move only there; hold the other stages as controls so seasonal footfall is not mistaken for a design effect.

Related

• Within the group: K7.01.1 Most passers-by never notice the display · K7.01.3 Static content almost never yields interaction
• Adjacent: K7.02 Discoverability of Interactivity · K7.03 Social Embarrassment · Z8.01 Ambient displays and peripheral information
• Search terms: audience funnel · passing by · focal awareness · public display engagement