Learn data science best practices

3 Features That Will Keep Your IoT Initiative From Failing

Despite the breakneck pace of the tech news cycle, it isn’t often that we in the technology sector come across news stories on technologies that are as transformational as the Internet of Things (IoT). There are many research reports and real-world evidence indicating that IoT will transform our personal, public, and vocational experiences. As of the summer of 2017, there were at least 450 IoT platform providers vying to win our connected experience-based business. IoT has also been catapulted into the mainstream by products that bridge the personal and professional worlds, such as Amazon’s Alexa, Google Home, etc.

The benefits of connected experiences are plenty when it comes to efficiency gains and automated business processes, as well as optimized decisions. However, in 2017, Cisco reported that 76% of IoT initiatives fail. Gartner, which was one of the first firms to project the creation of trillions of connected devices, didn’t classify any IoT solution provider as a leader in the ability to execute in its 2018 Magic Quadrant. So, obviously, there is something amiss in IoT adoption and execution.

What makes an IoT solution? 

A quick survey of the landscape of IoT solutions shows that these products and services broadly focus on three aspects, namely: 

  1. Devices: This includes protocols, controls, physical devices, etc.

  2. Data: Visualization of data and devices, including interactions via 2D, 3D, AR, VR, chatbot, etc.

  3. Automation: Analytical insights, optimal decisions, and automated business process flows

The majority of IoT solutions in the market today focus on devices and data. While connecting, visualizing, and interacting with digitized devices are part of IoT deployment, the goal of IoT isn’t to deliver those features. They simply add to our experiences. Put simply, IoT projects result in more things to visualize, more data to consume, and more decisions to make. While curiosity can help sell these features at the beginning of a project, no one wants to view thousands of devices, consume their data, and control them on the daily basis, even if it is via interactions with AR, VR, chatbot, etc.

What is the key deliverable from IoT?

The success of IoT projects lies in automation. That’s because automation is what contributes significantly to the return on investment of an IoT project. The promise of IoT is in self-optimization and the autonomy of connected things. We expect our connected things to automate our manual interactions and decisions. Interacting with those connected things should be rare, and when we do interact, it should be a frictionless experience via chatbot, AR, VR, mobile, digital twin, etc. So, a successful IoT project should include: 

  • Analytics to transform raw data into insights

  • A recommendation engine for generating optimal decisions

  • Integrations with appropriate applications to automate actions

Let’s take a look at each of these components.

Analytical Insights

Deriving insights from raw data spans a large spectrum of data science methods. Depending on the specific stage of a customer’s IoT journey, insights will need to be based on descriptive, exploratory, diagnostic, or predictive analytics. For a customer who is manually aggregating performance data, descriptive analytics that periodically calculate and report desired metrics are ideal. This can be addressed by algebra and statistics-based modules. 

If a customer wants to explore relationships between connected things and/or diagnose problems, however, the solution needs to include statistical and artificial intelligence (AI) or machine learning (ML) methods — such as anomaly detection, correlation, trend detection, segment analysis, or root cause analysis — to deliver specific insights. Furthermore, if a customer wants to forecast the future performance of connected things in order to avoid disruptions and optimize business outcomes, the IoT solution would require AI/ML-based predictive methods.

While a specific customer’s current stage of analytical maturity and the pace at which they progress may vary, it is no surprise that all IoT customers expect analytics to be part of the solution. Therefore, analytics in IoT solutions aren’t about a single problem or single AI/ML algorithm. Infusing analytics is about compiling the AI/ML-based insights a customer needs to derive from IoT in order to meet his or her business needs. 

Optimal Decisions

Analytical insights are usually the starting point of deriving value from an IoT project. Generally, insights require a human to assess their impact to business and implement the appropriate actions. Deriving appropriate actions, however, is typically the most time-consuming task. These actions can also vary based on the individual’s level of experience and usually requires optimization methods and reinforcement learning in addition to AI/ML algorithms.

For example, an analytical insight could be the predicted failure of an asset at future date. An optimal decision that will minimize the business disruption due to this predicted failure would be to recommend an alternate asset that can deliver the work of the failure-bound asset as well as a time window at which the failure-bound asset can be serviced. The user can then focus his or her energy on taking action, instead of figuring out how to derive the right action from insights.

The specifics of data science vary when the optimal decision to be made varies. Your target will change as business strategy changes. Therefore, the most desirable IoT solution has a comprehensive AI/ML offering that grows with the business.


The vast majority of IoT business benefits are derived from real-time analytics. However, a business will miss out on these benefits if real-time decisions cannot be executed without human attention. Therefore, the integration of IoT solutions with the appropriate business applications is key.

The value of IoT analytics lies in utilizing data from connected things in addition to contextual data from business applications. If they are not integrated with other applications, IoT analytics will be solely based on the data collected by connected things, ultimately providing a narrow view of business that results mainly in insights that require human judgement for action. Lack integration contributes to siloed IoT solutions, which naturally fail due to declining use beyond the initial period of adoption.

The business benefits of IoT do not come from a single deployment for one specific business problem. Instead, the benefits of IoT come from improving many business decisions. Therefore, the less-than-ideal state of IoT adoption today isn’t due to a lack of market readiness, but a lack of comprehensive IoT solutions. There are indeed solutions available that enable industrial companies to transform their businesses into digital powerhouses that deliver better employee, product, and customer experiences. These solutions have a significant impact because they deliver analytical insights, optimal decisions, and automated business processes. Therein lies the success of IoT. 

Be the first to comment

Comments ( 0 )
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.