The main reasons for failure during the proof of concept (POC) stage of IoT

• Sensing devices that measure environmental information and convert it into digital signals.
• Actuating devices that receive digital signals from the network and act upon them.

• Application layer. This layer is characterized by running embedded software, i.e., firmware or appropriate operating systems on sensing and actuating devices. It may also include mobile, web, and desktop applications that help users interpret sensor data and manage gadgets. Therefore, if a startup wants to create an IoT application, it may need to know that its application is just one of many IoT applications.
• Service and application support layer. Essentially, this is the IoT infrastructure layer where data aggregation, storage, and processing operations occur. To save costs and ensure uninterrupted device/service performance, IoT startups often opt to set up this infrastructure in the cloud (rather than on-premises servers).
• Network layer. IoT engineers can implement cellular, Wi-Fi, and wired connectivity technologies at the network layer to connect the components of the IoT ecosystem—i.e., the “things,” backend infrastructure, and user applications.
• Device layer. The capabilities enabled by the device layer can be divided into:

• Learning.
• Trial/Proof of Concept.
• Purchase.
• Adoption.

• IoT product idea validation.
• IoT product discovery.
• Minimum Viable Product (MVP) development.
• Market launch and MVP scaling.

• Assessing the demand for IoT solutions. In addition to learning from and studying analytical articles published by technology consulting firms like Gartner and Accenture, startups can conduct in-depth interviews with experts and potential customers in their target fields such as healthcare, wellness, manufacturing, retail, etc. Next, analyze macro and micro environmental factors affecting business using marketing frameworks such as TEMPLES, VRIO, and Porter’s Five Forces model. Special attention should be paid to managing data privacy and security laws regarding IoT adoption in the target geographic location, as well as industry-specific regulations for IoT devices, such as HIPAA, HL7, and NIST.
• Understanding competitors. As part of the macro environment audit, competitive analysis allows startups to determine the best feature set, pricing, and marketing strategies for their IoT products. The goal for startups is to identify a niche market and offer something that competitors lack—be it functionality, competitive pricing, superior quality, or five-star customer service.
• Selecting the right IoT business model. Leveraging insights gained from market and competitor research, startups should choose the right business model to monetize their IoT products. Some key options here include one-time purchases, subscriptions, and monetization of ancillary services and products, such as sensor data analytics. To better align service products with the company’s mission, resources, and marketing mix, Alexander Osterwalder’s Business Model Canvas template can be used.
• Estimating the workload required to build IoT devices. In this step, a SWOT analysis should be used to summarize market research findings and determine what resources and capabilities are lacking to create IoT devices and the applications supporting their logic. Then, based on the company’s main focus (hardware, embedded, web, or mobile), it will be determined which parts of IoT product development need to be outsourced.

• What tasks and processes will the IoT system enhance or automate?
• What type of data collection devices will be used?
• What connectivity technologies will the IoT product rely on?
• Where will sensor data be stored and analyzed? How will it be presented to the end user?
• How will the customized IoT solution interact with third-party devices and services?
• What is the approximate size of the target user base?

• Hardware design (plus certification).
• Infrastructure setup.
• Application development.

• Analysis. The analysis phase is primarily based on insights gathered by startups from the IoT product discovery, from concept development to technical requirements specifications.
• Modeling. Startups collaborate with hardware engineers and industrial designers to design printed circuit board (PCB) layout schemes and visualize the gadget’s casing in 3D CAD.
• Prototyping. Don’t confuse the IoT prototyping activities mentioned by startups in the discovery section with custom device prototyping. This time, off-the-shelf solutions like BeagleBoard, RaspberryPi, and other ready-made IoT development boards won’t be used. Instead, hardware manufacturers need to be contacted to produce up to ten PCBs based on the layout scheme created in the previous step. The hardware supplier for the startup will run extensive tests to verify that the PCB meets its performance requirements, debug it if necessary, and update the technical documentation.
• Testing. At this stage of the IoT product development lifecycle, engineers will convert successful prototypes into pre-production models while using different materials for the device casing. Next, the startup needs to conduct electrical safety, pre-certification, and user testing. If serious errors occur during this process, don’t be surprised. It is not uncommon for custom IoT devices developed by startups to take 6 months to 2 years to achieve performance and safety goals.
• Certification. During market research, startups have already learned about the IoT regulations effective in their target markets. However, depending on the application scope of their gadgets, various certifications may be required before selling the IoT solution to end users. These may include restrictions on hazardous substances (ROHS) and Energy Star compliance, certification from the Electrical Engineering Committee (EC) and Underwriters Laboratories (UL), Bluetooth SIG qualification licensing, and industry and product-specific testing certifications for gadgets that collect user data or come into direct contact with skin.

• Embedded software. Firmware, middleware, device drivers, and mature operating systems connect with the hardware components of their custom IoT devices, allowing them to perform intended sensing and actuating operations and helping integrate the gadgets with other devices and components of the IoT infrastructure. Typically, hardware suppliers working with startups can also handle embedded components, though a separate team may need to be hired for this.
• Connectivity. Similarly, the embedded team handles the networking part. To send sensor data to the gateway or directly to the cloud, the company’s gadgets will rely on short-range or long-range wireless connectivity technologies. When selecting a connectivity technology stack, network costs should be considered in advance. For instance, if cellular technology is chosen, every megabyte of data sent by the gadget over the network could ultimately cost $0.04.
• Cloud computing infrastructure. Based on the requirements determined during the discovery phase of the IoT product development lifecycle, startups need to choose a cloud platform that supports their gadget’s business logic. Here, sensor data will be aggregated, stored, analyzed, and visualized using dynamic dashboards. Cloud computing service providers like Google, Amazon, and Microsoft typically charge based on the number of server calls made by their gadgets or the number of devices in the IoT ecosystem. But the costs of cloud computing providers are not the only issue to consider here. When designing the blueprint architecture for IoT solutions, necessary preparations should be made for user base, data volume transmitted over the network, and overall system complexity. For instance, if a startup plans to deploy machine learning models in the future to interpret sensor data, it should be able to do so without a complete overhaul of the infrastructure. Device management, wireless (OTA) software updates, and ongoing performance optimization via DevOps are also important considerations.
• Support infrastructure. Establishing a data warehouse or data lake solution in the cloud platform and configuring some analytics functionalities is only half the job. Complex IoT solutions like remote patient monitoring (RPM) or end-to-end home automation systems require dedicated customer support departments and a multitude of relevant software tools, such as mobile, web, and desktop applications, to allow end-users and administrators to connect and operate devices.

• Native or cross-platform mobile applications that serve as remote controls for IoT products.
• Embedded human-machine interfaces (HMI) that allow users to operate devices without mobile or web applications.
• Web-based or desktop applications that reflect the functionalities of their mobile applications and allow IoT product administrators to manage user accounts.

• Form a beta user group to test the minimum viable product (MVP) and adjust the product (i.e., the application and the gadget itself) to better meet user needs.
• Develop a marketing plan that includes content creation, participation in specific industry and tech events, and building partnerships with influencers.
• After reaching initial revenue targets, gradually expand the product’s feature set by adding new functionalities and use cases.
• Double down on delivering an excellent customer experience: after all, the cost of acquiring a new customer is five times higher than that of retaining an existing one.

Finally, startups need to focus on the common challenges they face when developing IoT devices:

• Encountering technical barriers late in the IoT product development process. Creating a wristband with physical activity tracking features may seem like a great idea, but what happens if the metal casing interferes with Bluetooth signals, preventing the gadget from sending sensor data to the mobile application? A reliable way to avoid this is to kick off the project at the discovery stage before putting the device into production, ensuring extensive testing coverage.
• Struggling with multi-vendor IoT project management. Few companies have the expertise and personnel needed to build all components of an IoT system. As the owner of an IoT startup, one should elevate their project management knowledge, choose the right project tracking software, and keep their distributed hardware and software development teams aligned.
• Incorporating too many features into the IoT product MVP. Market research findings may indicate that users want a self-learning smart home system with biometric control options. In reality, startups are likely to lack the skills and resources to create such a complex IoT device from scratch (and in one iteration). It is advisable for companies to start their IoT product development journey by creating an MVP with sufficient features to spark user interest and attract investor involvement.
• Ignoring IoT scalability and hidden infrastructure costs. To select the right development technology stack and design an IoT solution architecture that can grow with the business, one should collaborate with skilled business analysts during the product discovery phase, engage with stakeholders inside and outside the company, and hire top-notch software architects.
• Neglecting IoT security. Despite efforts made by the global IT industry and governments, IoT remains an easy target for cybercriminals. From hardcoding device passwords to using open-source software development tools with file vulnerabilities, there are many ways to overlook security flaws in the IoT infrastructure, which will disappoint customers. This is why “designing for security” should become the mantra of IoT product development from day one.

51CTO Technology Community is recruiting

Today, your likes make me full of energy!