Placement is a variable, so treat it like one
Where a screen hangs decides most of what it earns, yet placement is usually settled once, during fit-out, by whoever holds the drawing. Retailers who A/B test everything on their website rarely test the single most expensive pixel real estate they own. The obstacle was never willingness; it was measurement. With zone-level flow data, placement becomes a testable variable like any other.
How do you A/B test digital signage placement?
Define the outcome first (dwell at the screen, visits to the promoted category, or sales of the promoted product), then vary one thing at a time: same content on two candidate positions (position test) or two content variants on one position (content test), never both at once. Measure passing traffic, dwell, and the downstream outcome for each variant over comparable periods, and let the baseline data decide how long the test runs. The measurement layer has to exist before the test: continuous counts and dwell at each position, plus zone visits for the promoted category.
Designing a placement test that survives scrutiny
Match the periods. Week-over-week retail traffic is seasonal, weather-driven, and promotion-driven. Compare like weekdays across equal-length windows, and keep a control zone (a category not being promoted) to catch store-wide swings that have nothing to do with the screen. The day-of-week effect alone is large enough to fake a result; the pattern is documented in day-of-week footfall.
Measure exposure, not just outcome. A position that wins on category visits but loses on passing traffic is telling you the content is strong and the position is weak. Without per-position traffic and dwell you cannot separate the two. Dwell measurement at screen positions is the core instrument here: digital signage dwell time.
One variable. Moving the screen and refreshing the content in the same week produces a number that means nothing. Sequence the tests.
What placement tests typically reveal
Three findings recur across venues. Eye-level positions along natural walking paths beat elevated positions that "cover" more area, because coverage is not attention. Positions near decision points (category entrances, queue lines) outperform positions near the entrance, where visitors are still in motion; the entrance-zone mechanics are the same ones described in the retail decompression zone. And dwell-heavy positions justify longer content loops, while pass-through positions need the message compressed to a glance.
The measurement infrastructure underneath
A placement test needs per-position passing counts, per-position dwell, and zone-level visit data for outcome measurement, running continuously so baselines exist before anyone asks for a test. Ariadne provides exactly that layer, camera-free: visitors register as counts, dwell, and paths, never as footage. For mall operators, the same infrastructure answers the adjacent question of where directory screens belong, covered in mall directory placement. For the outdoor version of this logic, the worked example is the billboard optimization case.
---
