AR is most defensible when it lets a visitor see information in relation to a real object, space, or action: a hidden layer, a before-and-after state, or a consequence that cannot be safely or economically built at full scale. If the content would work just as well as a video on a nearby screen, the camera, tracking, device, and onboarding requirements need a stronger reason to exist.
Test the reason for the overlay
Write down what the visitor should notice about the physical thing and what the digital layer adds. A model of internal machinery aligned to a real product may help explain a mechanism. A floating logo above a stand may add spectacle without helping the buying decision. This is a design test, not a published rule that AR always improves comprehension.
The empirical evidence is narrower than marketing claims often imply. Sommerauer and Müller ran a pretest–posttest crossover field experiment with 101 visitors at a mathematics exhibition and found better knowledge acquisition and short-term retention for augmented than non-augmented exhibits in that setting. In a separate 2018 crossover field experiment with 24 participants at a mathematics exhibition, they again found a short-term advantage, but it was not present at the two-month follow-up. These are two different samples studying mathematics learning, not product sales at an event. They support testing the actual knowledge outcome, including whether it lasts, rather than assuming a novel overlay creates lasting recall. Sommerauer and Müller, Computers & Education, 2014; Sommerauer and Müller, HICSS, 2018.
A Smithsonian case study of mobile AR in a natural-history museum examined common concerns such as distraction, replacement of real objects, gimmickry, and onboarding. The authors reported that several assumed concerns were not borne out in their case, while technical and social challenges remained. That is evidence from one museum project, not a blanket dismissal of those risks. Marques and Costello, Curator, 2018.
Compare AR with the simpler route
Make a small prototype of the AR moment and a non-AR version that communicates the same claim. Use the real venue light, device, object, and viewing distance. Ask first-time visitors to complete a task: identify the hidden component, compare states, or explain the process. Record whether they can start unaided, keep the physical object in view, understand the overlay, and finish without losing tracking.
The comparison should include the entire visitor path, not just the visual effect:
| Question | Why it matters |
|---|---|
| Who supplies and holds the device? | A phone requirement can exclude visitors or slow entry. |
| What does the camera point at? | The tracking target and bystanders affect the physical setup. |
| What happens when tracking fails? | A visitor needs a recoverable state, not a frozen overlay. |
| Can companions see the result? | A private handset may weaken group participation. |
| Is there an equivalent non-AR route? | The message should remain available when the technology cannot be used. |
These are commissioning questions, not a claim that every AR project needs the same hardware or footprint. The answer may be a projection, physical model, screen animation, or staff demonstration if it serves the task more reliably.
This decision differs from choosing gesture or touch input: AR is a choice about how digital information relates to the physical world, while gesture and touch are choices about control. Archv's interactive-experience service can evaluate both, beginning with the visitor's question and an honest prototype of the simplest way to answer it.