This study is unusual in that it follows robot deployment across the full project lifecycle — not just adoption intent or customer reactions, but the messy, practical work of actually getting robots running in a real restaurant environment. The researchers conducted 22 interviews with 34 participants across Norwegian restaurants and supplemented the interviews with direct observational fieldwork, producing one of the more grounded accounts of what robot implementation actually looks like on the ground.
The finding that will matter most to operators considering a first deployment: physical facility planning is decisive. Restaurants that attempted to introduce delivery robots into spaces not designed for them — narrow aisles, inconsistent flooring, poorly positioned service stations — faced expensive modifications and extended timelines. Those that incorporated robot requirements into early facility planning, even modestly, avoided these problems almost entirely. The implication is that robot deployment should be treated as a capital planning decision, not a technology procurement decision. If a remodel or new build is coming, that is the moment to design for robots. Retrofitting is always more expensive and always produces more friction.
On staffing, the results are more nuanced than either the optimist or pessimist narratives typically suggest. Robots did meaningfully reduce physical strain on wait staff — this was consistent across the restaurants studied and valued by employees who had previously dealt with the repetitive load of high-volume delivery. Where workplace inclusion was an explicit goal, robots enabled the hiring of staff who could not physically manage the full demands of traditional service roles, expanding the available labor pool in a sector that faces persistent shortages. However, robots also required staff retraining — not just technical orientation, but a deeper shift in how staff understood their own role. In restaurants where this mindset work was done deliberately, staff satisfaction was high. In restaurants where robots were introduced without it, the robots were underused and staff resentment was common.
For the customer experience, the novelty effect was real and durable across the study period, with robots generating positive reactions and social sharing. What the research does not support is relying on novelty as a strategic differentiator — as robot deployment becomes more widespread, the excitement will normalize. The properties that will maintain strong guest response are those that use robots to deliver measurably better service: faster delivery, more consistent timing, freed-up staff capacity for higher-value human interaction.
The project management dimension of the paper is worth reading carefully. The researchers identified initiation phase decisions — scope definition, vendor selection, facility assessment — as disproportionately important to long-run outcomes. Errors made in initiation were rarely fully corrected in later phases. For operators planning a first deployment, this argues for investing more heavily in the preparation phase than feels necessary, and less heavily in the technology evaluation itself. Most of the robot hardware available today is good enough; the differentiating variable is how well you have prepared your space, your team, and your service model to work with it.