For many technology enthusiasts, the Internet of Things (IoT) is the most important topic of the moment. They expect this technology to have an enormous impact on our lives. My personal experience over the past few years has led me to challenge several myths and unrealistic expectations surrounding IoT. Some of these myths contribute to mistaken assumptions, insufficient financial returns and disappointing results for all of us.

I hope my 12 observations will inspire you to use IoT for its intended purpose: creating business value.

Saying that IoT is about ‘things’ is like saying that office work is about using a keyboard. The ‘things’ are part of the process, but they are not the most important part. They are devices that generate data about the physical world. We need to process this data into meaningful information that forms the basis for decisions. Depending on the decision, an action is triggered to change something in the real world. When IoT is implemented correctly, that action happens automatically and creates real-world value. And that value is what matters. This could mean adjusting an electric motor’s settings to reduce energy consumption, proactively scheduling a building’s climate control to accommodate more visitors, or changing a food factory’s recipe to optimise flavour and nutritional value. An IoT implementation is only successful when you can link it to a use case that creates value for part of your organisation or community. IoT is therefore certainly not just about ‘things’: it is about generating value in the real world.

Machine-to-machine communication (M2M) dates back to the first SCADA and PLC systems. The ability to exchange data and commands between machines accelerated production and optimised operational processes. However, M2M is not the same as IoT. M2M is usually confined to an individual machine or factory. The Internet of Things makes data and commands available at a remote, cloud-based location, where they can interact with other data, metadata and logic. This allows us to compare data from different machines, monitor them remotely and optimise how machines and devices work together. M2M is a form of IoT, but lower bandwidth, computing and storage costs, enabled by the cloud, make IoT much broader, more sophisticated and more valuable than traditional M2M.

It is a common misconception that developing IoT applications is simply a more complicated form of software development. IoT requires a broader range of skills, and the data chain is much longer. Software developers assume that hardware, connectivity, software and storage are available. As an IoT developer, you need to account for sensors, embedded systems, different forms of mobile and wired connectivity, large-scale data processing logic, data storage, dashboards and integration with external systems. You also need to anticipate faults and failures throughout the IoT chain, while supporting fast, low-latency data processing, security and privacy. Sometimes you have to do this with very limited resources. At the edge or sensor level in particular, you are constrained by the hardware’s physical capabilities: you cannot simply add memory or scale up computing resources. Generally, there is no single, generic IoT developer role because the expertise required is so diverse. You typically see engineers specialising in embedded systems, cloud data processing, dashboards or device management. An IoT developer is more than an advanced software developer: the role also involves hardware expertise, connectivity engineering, performance engineering, security, user interface design and much more.

IoT is often associated with big data. The concept of big data rests on the idea that patterns and relationships will emerge automatically once you have enough data. With the arrival of IoT devices, data volumes are exploding. Data is collected automatically, and sensors sample data points that are not needed to support a business case. But who cares? We have a data lake to store all that excess data. Apply some algorithms or machine learning, and countless patterns will emerge, right?

This thinking is flawed and highly inefficient. Being able to measure large numbers of data points does not mean you should. All these data points obscure the business case and muddy the data lake. IoT should always focus on business results and generate data that serves its intended purpose.

The most common misconception about IoT is that it is simple: use a basic development kit and you will be up and running within a few hours. This is precisely why so many IoT projects stall in their initial stages. You can build something simple very quickly, but things become complicated as soon as you want to scale. You need to manage multiple devices, process their data and implement privacy and security measures. When something goes wrong, and believe me, something will, you need monitoring to detect faults, followed by investigation and resolution. As the application grows further, you need to organise device provisioning, data processing, data storage, cloud management and cost management. Before long, you reach a point where you can no longer shut down the IoT solution for maintenance. The capabilities it provides have become so essential to operations that the solution is business-critical and must remain available at all times. Think of a train that a technical team has to configure and improve while it is travelling at full speed. The greatest challenge of IoT is therefore being prepared from the outset for an environment that grows and changes rapidly.

IoT enthusiasts are always talking about smart sensors, smart meters and smart devices. This description is not entirely accurate. In most cases, ‘smart’ simply means that a device is digital and connected. But, to be honest, these ‘things’ are not smart. The connected software is sometimes smart, but the smartest components are in the cloud, where logic processes data and turns it into conclusions and actions. The cloud is where engineers apply machine learning and develop algorithms. Actions, instructions or algorithms are then sent back to the ‘things’ to make them appear smart. But the smartest part of IoT happens in the cloud.

It is a common misconception that sensors and actuators require no maintenance after installation. At home, manually updating ten sensors is relatively straightforward. Managing twenty thousand sensors in a professional environment is a very different matter. Sensors have an average lifespan of five to ten years. During that period, a critical security update, a settings change or an adjustment to the logic is highly likely to be needed. And that is before we even consider replacing batteries and other physical maintenance. Visiting sensors at their physical locations is the most expensive activity in IoT. It can make your business case unviable. Planning for remote maintenance, monitoring, updates and patching is therefore essential for IoT solutions. This process must be watertight, resilient and highly reliable.

Once devices are connected, people assume they can also communicate with one another. In most cases, this is neither true nor necessary. For direct data exchange between devices, we see the following possibilities:

• Devices from the same supplier are usually interconnected. In home automation particularly, using the same remote control or app makes monitoring and controlling these devices easier.
• Devices using the same type of protocol may be able to connect to one another.
• You can connect devices when a use case requires it and doing so creates value.
• Connecting devices from entirely different domains serves no purpose. I cannot think of a use case, for example, in which your smart car needs to tell your smart TV that you are coming home.

The market remains highly competitive and fragmented, so a universal communication protocol is still a long way off. The only exception at present is home automation, where a common communication protocol supported by different hardware suppliers is emerging. The diversity of use cases does not actually require all devices to be interconnected. Ultimately, they are logically grouped by domain. Communication standards are defined within each domain, and communication protocols follow from them. The idea that all IoT devices will work together is an illusion.

The whole purpose of IoT is to process device data at scale. We use cloud technology to enable this because it is the most effective way to add resources, such as computing power and memory, dynamically to applications. This does not mean that scaling an IoT solution requires no planning. Unfortunately, there is no magic dial you can turn to make an IoT project with ten devices suddenly process data from ten thousand.

Scaling IoT presents a range of challenges, with different optimal approaches and tipping points depending on the number of devices, reliability requirements and total volume of data processed. These tipping points can be described as paradigm shifts. You use a particular solution for a specific task until you reach a certain scale. Beyond that point, continuing in the same way becomes almost impossible, and you need an entirely different approach or design. Take device numbers: manually registering and provisioning a hundred devices is manageable. With a thousand, it becomes challenging, and you will want to prepare registration and provisioning in batches. With ten thousand devices, you will want to implement self-registration and automated provisioning.

The same applies to data processing. As you scale, you need different, more robust components to ingest, store and process data. Some components are simply too expensive for a small number of messages but become essential as device numbers grow. Message queuing, parallel processing, caching and automated message clean-up before processing then become important. Without the right expertise, scaling can cause your cloud costs to grow exponentially, undermining your business case. When scaling IoT solutions, you must always anticipate the next tipping point and design the next approach for transferring IoT data at scale. Only this approach prevents your IoT solution from becoming congested and unusable.

Conventional IT security is the ‘junior league’ compared with the security required for IoT devices. The ‘things’ are left unattended in the field: on a truck, in an office building, in a public space, in a factory or in a home. They may be connected to the core of your organisation’s data processing and decision-making systems. This gives malicious actors unlimited time to examine, open, modify or replace them with something that could harm your organisation. The consequences could include manipulated data, hacking attempts, denial-of-service attacks or the transformation of your edge devices into a digital zombie army completely beyond your control. When it comes to IoT security, you must assume that your ‘things’ will be compromised. When that happens, you need to assess the impact quickly and find a solution. While a security issue in IT systems may lead to data loss, an IoT security issue can create dangerous situations in cities, factories, traffic and energy networks. Design with safety and security in mind, particularly when using actuators to change the state of ‘things’ in the real world.

IoT is often seen as something that belongs in an innovation department, where specialists tinker with sensors, circuits and cloud components. Unless you work for a large original equipment manufacturer (OEM) or a research organisation, however, most IoT projects are not about technological innovation. They are about business model innovation enabled by data. In most cases, IoT provides the foundation for a new business model or industry innovation using devices and data. The technology can be complicated, but the real challenge and innovation lie in changing business models and organisations.

Some managers believe that legacy assets are unsuitable for integration into their digital transformation. However, implementing IoT entirely from scratch in a greenfield environment is an illusion. Most, if not all, projects start in a brownfield environment. You connect and optimise existing asset infrastructure using data, while continually adding new digital devices. Adding sensors to existing equipment can generate valuable information throughout the organisation: on the operational floor, across the value chain, in optimisation programmes and in management reporting. Simply adding a motion, energy or temperature sensor to older equipment can generate enough data to compare machine performance, identify differences in energy efficiency or plan maintenance based on actual machine usage.

I imagine there are many more IoT myths to debunk. I am always open to discussing them, so feel free to get in touch. The main conclusion is that IoT is a serious field that requires serious expertise to succeed. Its driving force should always be the value it delivers to people, organisations or society as a whole.