A physical product token can make an exhibit inviting: pick it up, place it on a table and watch the story change. A touchscreen can offer the same content without objects to make, calibrate, clean or lose. The right question is not which interface feels more futuristic. It is what the visitor must understand or compare, and whether handling an object helps them get there.
The ADNOC tangible installation described by Ultratec used physical tags to trigger related content. B·U·T's Nakheel project paired object recognition with a table for property discovery. Those are useful examples of what was built, not controlled proof that object recognition outperforms touch for another buyer's products.
Decide what the object contributes
Use a physical object when its shape, material or relationship to the real product is part of the explanation. A sample tile could let a visitor compare finishes; a recognizable miniature could anchor an otherwise abstract system. If the object is merely a round button with a logo, a clear on-screen control may deliver the same choice with less operational work.
Map the visitor's task before selecting the input. Can someone find two relevant products, see the meaningful difference, and leave with a next step? Does the table need to handle several people at once? Is the content updated weekly? How will staff replace or re-pair a missing token? These are procurement questions, not universal rules.
The evidence argues for a specific, bounded advantage. In an Exploratorium field comparison of a phytoplankton-data exhibit, researchers observed 253 visitors across physical-ring and virtual-ring versions, plus 75 recruited think-aloud pairs. Physical rings invited more touching and group use. The comparison did not detect more thorough interaction, more relevant questions or more on-topic talk. That is one museum exhibit, not a product-sales study. A separate 2016 laboratory study of 20 university participants reported faster completion of a genomic data-visualization task with tangible controls because participants used a more effective exploration strategy. It does not establish a general conversion or comprehension benefit for product showrooms.
Run a fair pilot before specifying the final table
Build two low-cost versions with the same product facts and comparison steps. Version A uses the proposed object and a simple reader. Version B uses visible touch controls. Keep the screen, labels, content, distance and staff script as close as practical. If the tangible version uses better photography or a shorter flow, the result will tell you little about the input method.
Ask new users to complete the actual decision task without coaching. Record the first attempted action, whether they discover how to start, whether they complete a useful comparison, what they say the difference is, and where they request help. Invite pairs as well as solo users if the final setting expects groups. Observe reset between turns. A measure such as time at the table is useful diagnostic context but is not proof of understanding.
| If the pilot shows… | Procurement implication |
|---|---|
| More people handle the object, but comparisons are no clearer | Keep the object only if attraction or shared play is an explicit goal worth its operating cost. |
| The object helps people explain a material or spatial difference | Specify the object, tracking tolerance and replacement procedure. |
| The object creates wrong triggers or blocks another visitor | Revise the sensor layout or use touch for that task. |
| Both versions work similarly | Prefer the simpler operational option unless the physical form carries a brand or product need. |
These are decision rules for a pilot, not results of an Archv experiment. Agree what counts as a correct comparison before watching users; otherwise teams can reinterpret every hesitation as success.
Budget for the complete operating system
The quote should name object fabrication, sensing, spare units, storage, cleaning, calibration, content update path, failure state and an operator's restart steps. If an object disappears, the visitor should still get a truthful route to the content. If a sensor reads two objects at once, the screen should resolve the ambiguity instead of silently showing the wrong product.
Archv's gesture-versus-touch guide addresses a different choice: contact-free input versus direct screen control. For a tangible table, the distinctive decision is whether the physical product cue itself improves the discovery task. Test that claim with people and content similar to the real installation before committing the build.