\u201cTwo years ago, predictive maintenance was forecast to be one of the most promising uses of the industrial Internet of Things (IoT).\u201d\nThat\u2019s the lead of report based on a recent Bain & Company survey of more than 600 high-tech executives (Beyond Proofs of Concept: Scaling the Industrial IoT, by Bain partners Michael Schallehn, Christopher Schorling, Peter Bowen and Oliver Straehle). The report goes on to note that identifying precisely when equipment might fail \u201cseemed like a no-brainer.\u201d And yet, the report concludes, \u201cpredictive maintenance has failed to take off as broadly as expected.\u201d In fact, industrial leaders were not as excited about predictive maintenance as they were back in a 2016 survey.\n\nPredictive maintenance hard to implement, hard to derive value\nAccording to Bain, there have been problems on the both sides of the ball: First, implementing predictive maintenance has been harder than expected, and second, deriving valuable insights from the data gathered has also turned out to be unexpectedly challenging.\nBut wait, it gets worse.\nIt seems that \u201cpredictive maintenance is just one of many IoT use cases that customers have had difficulty integrating into their existing operational technology and IT systems.\u201d While investment in proof-of-concept projects continues, the Bain report said, actually turning that into successful mainstream implementations hasn\u2019t been able to keep up. Long-term enthusiasm for the technology remains strong, the survey showed, but many industrial organizations now foresee implementation taking longer than initially predicted. (Bain still predicts the industrial IoT market to double by 2021, topping $200 billion.)\nAn evolution in industrial IoT concerns\nThose delays come along with a shift in the concerns of industrial IoT users. In 2016, top concerns centered on security, return on investment, and integrating IoT solutions with existing IT systems. After a couple years of PoC projects, security remains a key concern in 2018, and worries around gathering sufficient technical expertise, dealing with multiple data formats, and transition risks have become even more intense. ROI concerns, meanwhile, have faded a bit. (Could it be that industrial IoT projects are actually paying off?)\nNone of this should come as a huge surprise. The transition from promising vision to practical success often faces unforeseen roadblocks. But while these impediments can loom large during the early adoption phase, they tend to be quickly forgotten once the technology achieves widespread success, Indeed, the Bain report notes that \u201cgiven the progress in sensor technology, 5G connectivity, edge computing and edge analytics, and an estimated 20 billion devices connected by 2020, there\u2019s little doubt of the potential for technology to vastly improve efficiency and no doubt that the IoT will have to manage it.\u201d\nSuggestions for IoT vendors\nTo overcome these issues, Bain says, analytics firms, industrial technology makers and cloud service providers must help their customers gain deeper experience in industry-specific applications and offer more complete, end-to-end IoT solutions.\nIn the short run, though, that may not be so easy, for a variety of reasons.\nFirst, industrial IoT vendors and users don\u2019t always agree on what\u2019s important and what\u2019s ready for prime time, the Bain survey revealed. For example, while quality control and remote monitoring and tracking of equipment ranked high among both groups, customers wanted augmented reality\/virtual reality (AR\/VR) and energy management solutions, which vendors were less ready to deliver. On the other hand, vendors said they were ready to deploy predictive maintenance solutions, but it\u2019s not clear customers are waiting in line to buy them. (Notably, smart-city use cases, which have received a lot of press attention, were not top of the list for either vendors or customers.)\nSoftware deficiencies also remain a problem. According to Bain, \u201cDevice makers and other vendors of industrial and operational technology need to dramatically improve their software capabilities\u2014not a historical strength for most of them.\u201d The report notes that many vendors are spending freely on acquisitions to accumulate the required capabilities.\nFinally, Bain suggests that industrial IoT vendors focus on enabling key use cases and overcoming critical barriers, plus leverage partnerships to fill the inevitable capability gaps.