Your Ultimate Guide to Custom IoT Product Development

[UPDATED in May 2026]

Surveys consistently show that 75–80% of IoT projects fail to reach successful deployment. And while Beecham Research notes a 28% improvement in IoT project success metrics since 2020, only roughly one in four initiatives today is considered an unqualified success.

The reasons vary. Microsoft's IoT Signals research points to high scalability costs, technical challenges, and vague ROI as the leading causes of project failure at the proof-of-concept stage — as illustrated in the chart below. A separate 2025 study by Eseye adds a hardware-specific dimension: three in four failed IoT projects were undermined by poor hardware design, with 66% of businesses reporting that their IoT devices fail to connect regularly due to hardware issues.

High scalability costs top the list of custom IoT product development challenges. Source: Microsoft

As a startup considering IoT solution development, you can avoid the majority of these challenges by carefully planning your IoT pilot beforehand.

Here's where our guide to custom IoT product development comes in handy.

Understanding Custom IoT Product Development

To help you build an IoT device and the accompanying software ecosystem in a risk-free way, we're starting an article series that dives into the Internet of Things technologies and IoT product development best practices. Check out our blog to better navigate the dynamic IoT technology landscape!

This time, we'll focus on the Internet of Things definition, architecture, and stages your connected product goes through before hitting the shelves.

What Is an IoT Product?

The Internet of Things (IoT) is a network of physical objects connected to the Internet and/or each other over a wired or wireless network.

The "things" term may apply to electronic devices, such as fitness trackers, and non-electronic objects enhanced with sensors and lightweight control gadgets (think smart curtains operated via a relay and mobile app).

There are two types of IoT solutions you could create:

• Sensing devices, which measure information on the surrounding environment and convert it into digital signals
• Actuating devices, which receive digital signals from the network and act upon them

These devices can talk to the nodes within an IoT ecosystem (i.e., peer-to-peer communication), connect to the network via a gateway, or establish gateway-less connections.

What Does an IoT Solution Architecture Look Like?

Our custom IoT product development guide would not be complete without an IoT reference architecture. The diagram above includes an intermediary edge layer that accumulates data for immediate analysis. Source: ITU-T

For companies looking to tap into custom IoT product development, it is essential to understand how connected solutions function under the hood.

The Internet of Things architecture comprises four levels:

• Application layer. This layer features embedded software—i.e., firmware or proper operating systems—that runs on sensing and actuating devices. It may also include mobile, web, and desktop applications helping users interpret sensor data and manage gadgets.
• Service and application support layer. 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 choose cloud-based infrastructure.
• Network layer. On the network level, IoT engineers can implement cellular, Wi-Fi, and wired connectivity technologies to interface the components of an IoT ecosystem.
• Device layer. Includes gateway capabilities (supporting IoT communication protocols like Bluetooth, Zigbee, Z-Wave, and LPWANs) and regular device capabilities (data collection, edge processing, sleep/wake modes).

In 2026, the device layer increasingly includes on-device AI inference capabilities—also called edge AI or TinyML—where machine learning models run directly on the microcontroller rather than in the cloud. This adds a fifth functional consideration to architecture design, covered in detail below.

The Internet of Things architecture also incorporates device management and security components. Commonly, this functionality is baked into popular IoT platforms, such as AWS IoT Core, Microsoft Azure IoT Hub, and purpose-built alternatives like Particle or Losant.

What Stages Does the Custom IoT Product Development Lifecycle Span?

Prominent IoT infrastructure vendors like Microsoft and Google distinguish four stages of the IoT development process:

• Learn
• Trial/proof of concept
• Purchase
• Use

Here at Expanice, we prefer a slightly different classification, which better aligns with the stages IoT product startups actually go through:

• IoT product idea validation
• IoT product discovery
• Minimum viable product (MVP) development
• Market launch and MVP scaling

Let's inspect the activities undertaken during these phases of the custom IoT product development lifecycle.

IoT Product Idea Validation

A key step in IoT product development is validating whether there is market demand for your solution and a suitable monetization strategy to scale it post-launch

The global IoT market is projected to expand from $547 billion in 2025 to $865 billion by 2030, with the connected IoT devices installed base forecast to reach 39 billion by 2030. Despite this growth, most IoT startups will not capitalize on it without a disciplined idea validation process.

To build a product with strong commercial appeal, your startup should start your custom IoT product development journey with thorough market research. Its elements include:

• Assessing the demand for your IoT solution. Besides studying research papers issued by technology consulting companies like Gartner and Accenture, your startup could conduct in-depth interviews with experts and potential customers from your target domain—e.g., healthcare, wellness, manufacturing, retail, etc. Next, analyze the macro- and micro-environmental factors affecting your business using marketing frameworks like TEMPLES, VRIO, and Porter's Five Forces. Pay special attention to data privacy laws and, as of 2026, new IoT-specific cybersecurity regulations in your target markets. The EU's Radio Equipment Directive (RED) cybersecurity requirements became mandatory in August 2025, and the EU Cyber Resilience Act (CRA) begins its phased application in September 2026, with full compliance required by December 2027. Any connected product sold into European markets must be designed with these obligations in mind from day one.
• Getting to know your competition. Competitive analysis allows you to determine the optimum feature set, pricing, and marketing strategy for your IoT product. Your goal is to identify an underserved niche and offer something your competitors are missing.
• Choosing a suitable IoT business model. Drawing on the insights from the market and competitor research, your company should choose an appropriate business model to monetize your IoT product. Some popular options include one-time purchases, subscriptions, and the monetization of accompanying services and products, such as sensor data analysis. To better align your service offering with your company's mission, resources, and marketing mix, you could use the Business Model Canvas template by Alexander Osterwalder.
• Estimating the efforts required to build an IoT device. In this step, you need to summarize your market research findings using the SWOT analysis and determine what resources and capabilities you lack to create an IoT device and the applications supporting its logic. Based on your company's primary focus (hardware, embedded, web, or mobile), you'll figure out which parts of custom IoT product development need to be outsourced.

IoT Product Discovery

A discovery phase helps IoT startups identify technology roadblocks early on, increasing product success rates

In the IoT product development lifecycle, the discovery phase helps verify your IoT product idea against your business needs, evaluate your project scope, and create a preliminary technical vision for your custom IoT solution.

To reach these objectives, enlist the help of a skilled business analyst and solution architect. The IT specialists will collaborate with your company's internal and external stakeholders and determine what the IoT solution should do and how it is supposed to function.

These characteristics are known as functional and non-functional requirements for IoT product development.

Following the discovery phase, you'll get definite answers to such questions as:

• What tasks and processes would your IoT system enhance or automate?
• What type of data capturing devices are you going to use?
• Which connectivity technologies will your IoT product rely on?
• Where will sensor data be stored and analyzed, and how will it be presented to end users?
• How will your custom IoT solution interface with third-party devices and services?
• What is the approximate size of the user base you're targeting?

The discovery phase of your IoT product development project should also address edge AI requirements upfront: will your solution need to run inference on-device? If so, hardware selection—specifically, choosing a chip with sufficient on-chip compute and memory for TinyML workloads—must happen here, not after prototyping. Retrofitting edge AI capability onto underpowered hardware is one of the most expensive mistakes a connected product team can make.

Based on this information, you'll be able to select a suitable technology stack, lay the foundation for an IoT architecture that would flexibly scale, and get a realistic IoT cost estimate.

IoT Prototyping

Using off-the-shelf IoT development boards is a sure-fire way to test your Internet of Things idea without breaking the bank

The goal of the prototyping phase of the IoT product development lifecycle is to create a proof-of-concept version of your connected device, identify technology roadblocks, and test the prototype with real users to further refine its functional and non-functional requirements.

One of our customers, for instance, wanted to create a smart home security system based on motion sensors. These sensors were supposed to track movement both inside and outside residential buildings. During the discovery phase, our IoT product development team came to realize that the ratio between the measured data properties prevented the software from timely notifying users of suspicious activities. As a result, we replaced the sensors with Wi-Fi-enabled video cameras.

To build an IoT device prototype, you (or the vendor of your choice!) could leverage off-the-shelf single-board computers and microcontrollers like Arduino Uno and Raspberry Pi. In 2026, newer prototyping platforms—such as the Arduino Nano 33 BLE Sense and the Raspberry Pi 5—include on-chip ML inference support, allowing startups to prototype edge AI features without custom silicon.

The choice of ready-made IoT prototyping tools is based on initial hardware requirements, such as connectivity, power consumption, RAM and flash memory, system architecture, and the availability of SDKs.

The benefits of prototyping in IoT are mostly cost-related. For example, you could create a working version of your connected solution at a small fraction of what it would cost you to design a custom device. Also, you can start developing firmware, back-end infrastructure, and mobile apps faster and rule out technology limitations early on.

IoT MVP Development

As mentioned earlier, custom IoT product development is not limited to creating electronic devices. If you're working on an asset-tracking solution based on BLE beacons, you don't have to design custom hardware and could focus solely on building a supporting software infrastructure.

Otherwise, MVP development for IoT would span three stages:

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

Custom IoT Hardware Development

Custom hardware development typically consumes between 20% and 40% of an IoT product development budget

How to create an IoT device? Well, pretty much like the other IoT product development activities, the custom hardware design process involves several steps:

• Analysis. From concept development to technical requirements specification, the analysis phase largely builds on the insights you've gleaned from IoT product discovery.
• Modeling. You collaborate with hardware engineers and industrial designers to devise printed circuit board (PCB) layout schemes and visualize the gadget's case in 3D CAD.
• Prototyping. Do not confuse the IoT prototyping activities described in the discovery section with custom device prototyping. You won't be using BeagleBoard, Raspberry Pi, and other off-the-shelf IoT development boards this time. Instead, you need to contact a hardware manufacturing company and produce up to ten PCBs based on the layout schemes created in the previous step. Your hardware vendor will run extensive tests to validate that the PCBs meet your performance requirements, debug them if necessary, and update the technical documentation.
• Testing. At this stage of the IoT product development lifecycle, engineers will transform successful prototypes into pre-production models while using different materials for the device case. Next, you'll need to conduct electrical safety, pre-certification, and user tests. Do not be surprised if critical errors surface in the process. This is not uncommon, and this stage of IoT product development can last anything between six months and two years before you achieve your performance and safety goals.
• Certification. When doing market research, you have learned about the Internet of Things regulations that are effective in your target markets. Depending on your gadget's scope of application, you might need to procure various certificates before selling the IoT solution to end users. These may include the Restriction of Hazardous Substances (ROHS) and Energy Star compliance, the Electrotechnical Commission (EC) and Underwriters Laboratories (UL) certifications, Bluetooth Sig Qualification clearance, as well as industry- and product-specific test certificates for gadgets that collect user data or come in direct contact with the skin. If you're planning to sell your device in the EU, add RED and CRA to the list.

IoT Infrastructure Development and Setup

The Internet of Things solution infrastructure term can mean a multitude of things, from a back end giving voice to your connected device to a dedicated customer department ready to solve user problems 24/7

The infrastructure layer of an IoT system includes several components:

• Embedded software. Firmware, middleware, device drivers, and full-fledged operating systems interface the hardware components of your custom IoT device, allow it to perform its intended sensing and actuating operations, and help integrate the gadget with other devices and components of an IoT infrastructure. Typically, the hardware vendor you're working with can handle the embedded part, although you might need to hire a separate team.
• Connectivity. Again, it's your embedded team that tackles the networking part. Your gadget will rely on short-range or long-range wireless connectivity technologies to send sensor data to a gateway or directly to the cloud. In 2026, two connectivity developments are worth attention: Wi-Fi HaLow (802.11ah) is gaining traction for low-power, long-range IoT applications, and the Matter 1.5 standard (published November 2025) has become the default interoperability layer for new smart home and building IoT products.
• Cloud infrastructure. Based on requirements from the IoT product development discovery phase, select a cloud platform that supports your gadget's business logic. The two dominant native IoT cloud suites in 2026 are AWS IoT Core and Azure IoT Hub, both of which typically charge per device or message volume. Note that Google discontinued its native IoT Core service in August 2023 and has not replaced it with an equivalent—startups evaluating GCP should factor this in before committing to that ecosystem. When designing your architecture, account for user base growth, data volume, and future requirements—including machine learning model deployment—so you don't need a complete infrastructure overhaul later.
• Edge computing layer. Edge computing has moved beyond experimentation into production deployment for IoT. An edge layer—sitting between devices and the cloud—processes latency-sensitive data locally, reduces bandwidth costs, and enables real-time decision-making where cloud round-trips are impractical. For startups, the key architectural decision is determining which workloads belong at the edge versus the cloud and designing your IoT infrastructure to support both from the outset.
• Supporting infrastructure. Setting up a data warehouse or data lake/lakehouse solution in the cloud and configuring some analytics capabilities is only half the job. Complex IoT solutions like remote patient monitoring (RPM) or end-to-end home automation systems require a dedicated customer support department and a plethora of related software tools like mobile, web, and desktop apps enabling end users and admins to operate connected devices. As part of custom IoT product development, you must address these issues beforehand.

Edge AI and TinyML: What IoT Product Teams Need to Know in 2026

Adding edge AI to your IoT product functionality can become a critical differentiator in healthcare, manufacturing, automotive, and energy and utilities

One of the most consequential shifts in IoT product development over the past two years is the mainstreaming of on-device AI—also called edge AI or TinyML. In 2026, the question has moved from 'Can we run AI on a microcontroller?' to 'When does it make commercial and operational sense to do so?'

The short answer: when your product needs to make real-time decisions faster than a cloud infrastructure allows, when connectivity is intermittent or expensive, or when raw sensor data must not leave the device for privacy or regulatory reasons. Predictive maintenance sensors, anomaly detection in industrial equipment, keyword spotting in voice interfaces, and gesture recognition in wearables are all being deployed with edge inference today.

What's changed technically: new low-power silicon—including Arm Cortex-M55 cores with Helium DSP extensions and RISC-V chips with integrated NPUs—now makes it practical to run quantized ML models within 256 KB–2 MB of RAM on coin-cell-powered devices. Development toolchains like Edge Impulse, TensorFlow Lite Micro, and ExecuTorch have matured significantly, though embedded MLOps (model versioning, OTA model updates, and drift monitoring) remains less standardized than cloud ML pipelines.

Implications for the IoT product development process:

• Discovery phase. Decide early whether your product requires on-device inference. Hardware with insufficient compute headroom for your target model cannot be upgraded later.
• Hardware design. Evaluate MCUs not just on power consumption and connectivity but also on energy per inference, memory capacity, and NPU availability. Platforms like Nordic nRF9151, Silicon Labs BG27, and STM32N6 are increasingly common choices for AI-capable IoT designs.
• Infrastructure design. Plan for OTA model updates from day one. Your inference models will drift over time as real-world conditions diverge from training data. The ability to push updated models to deployed devices without a full firmware OTA is a significant operational advantage.
• Testing. Add model accuracy and inference latency to your hardware test matrix alongside standard functional and electrical tests.

The Internet of Things App Development

Being part of the IoT product infrastructure, user-facing applications help configure and manage connected devices and visualize sensor data processed in the cloud.

Depending on your overall IoT product requirements and target audience, you might need to create:

• Native or cross-platform mobile applications, which act as a remote control for IoT products
• Embedded human-machine interfaces (HMIs) that allow users to operate devices without a mobile or web app
• Desktop or web applications that mirror the functionality of their mobile counterparts and allow IoT product admins to manage user accounts

Speaking of MVP development for IoT, it normally takes three to six months to create a complete software ecosystem for an IoT device.

The good news is that software development activities can run in parallel with custom hardware design. And if you're developing IoT devices on a shoestring budget in the hope of getting funded, you can skip the hardware design part altogether and make do with an off-the-shelf board for the time being.

For example, MedAngel, a healthcare technology startup from Germany, came up with an idea of a temperature-tracking device for insulin. The company chose the WunderBar platform as their primary custom IoT product development tech stack, placed the IoT board inside a keyring-like plastic case, and built simple mobile apps for sensor data interpretation. With an MVP on their hands, the MedAngel team got media coverage, took part in several tech contests, and launched a successful Indiegogo campaign. The company then scaled the IoT product's use cases across other temperature-sensitive medications and revamped the gadget's design.

IoT Product Launch and Scaling

Congratulations! You've built your first connected device containing enough features to meet user expectations and differentiate your company from the competition. Now it's time to get your IoT solution to the market, analyze initial user feedback, and tweak your product accordingly.

Technically, going to market is not part of the IoT product development lifecycle; that is why IoT first-timers often overlook it. Here's what you can do to avoid getting stuck in the IoT startup limbo:

• Assemble a beta user group to test your MVP and adjust your product—i.e., the applications and the gadget itself—to better meet user needs
• Develop a marketing plan covering content production, participation in industry-specific and technology events, and partnerships with influencers
• Gradually expand your product's feature set by adding new functionality and use cases once you hit your initial revenue targets
• Go the extra mile to provide superior customer experience: after all, acquiring new customers costs five times as much as keeping existing ones

Five Ways Your IoT Product Development Project Could Go Awry

From overstuffing your MVP with features to ignoring the innate Internet of Things security issues, there are many ways your custom IoT product development project could go off the right track

Finally, we'd like to draw your attention to common challenges startups face when developing IoT devices:

• Stumbling upon technology roadblocks late in the IoT product development process. Creating a luxurious gold bracelet with physical activity tracking capabilities might be a good idea, but what if the metal case interferes with the Bluetooth signal, preventing the gadget from sending sensor data to a mobile app? A surefire way to avoid such scenarios is to kick your project off with an IoT product discovery phase and ensure extensive test coverage before sending the device to production.
• Struggling with multi-vendor IoT project management. Few companies possess the required IoT product development expertise and personnel to build all components of an IoT system under one roof. As an IoT startup owner, you should elevate your project management knowledge and choose appropriate project tracking software to keep your distributed teams involved in hardware and software development on the same page.
• Incorporating too many features into an IoT MVP. The results of your market research might indicate that users want a self-learning smart home system with biometric control options. In reality, you most likely lack the skills and resources to create such a complex IoT device from the ground up (and within a single iteration). At Expanice, we recommend that our clients start their custom IoT product development journey by creating an MVP containing just enough features to ignite user interest and get investors on board.
• Ignoring IoT scalability and hidden infrastructure costs. To choose the right set of IoT-enabling technologies and design an IoT solution architecture that would grow along with your business, you should partner with a skilled business analyst during the product discovery phase. Additionally, you should interview stakeholders both within and outside your company and hire top-notch software architects, no matter the price.
• Taking IoT security lightly. Despite the global IT community's and governments' efforts, the Internet of Things remains a low-hanging fruit for cybercriminals. From hard-coding device passwords to using open-source software development tools containing documented vulnerabilities, there are a million ways to overlook security loopholes in your IoT infrastructure—and let your customers down. That's why "security by design" should be your IoT product development mantra from day one.

Stay tuned for more—and follow Expanice on social media to not miss industry insights and expert takes on novel technologies disrupting the Internet of Things market! Or drop us a line to discuss your upcoming or ongoing IoT project.

FAQs: Navigating IoT Product Development

1. How does edge AI change the IoT product development lifecycle?

Edge AI—running machine learning models directly on the IoT device rather than in the cloud—affects every stage of the IoT product development process in 2026. During discovery, you need to determine upfront whether your use case requires on-device inference, because this influences your hardware selection. During hardware design, you'll evaluate MCUs on compute headroom and NPU availability alongside traditional criteria like power consumption and connectivity. During testing, you'll add model accuracy and inference latency to your test matrix. And throughout the product lifecycle, you'll need an OTA model update capability to retrain and redeploy models as real-world data drifts from your training set. The payoff: products that make real-time local decisions, operate reliably offline, and handle latency-sensitive applications that cloud-dependent architectures cannot serve.

2. What effect does 5G have on IoT product development?

The 5G rollout continues to mature the infrastructure for ultra-fast, low-latency IoT communication. Global cellular IoT connections are expected to approach 8 billion by 2030, driven in part by 5G's expanding role in high-reliability, low-latency use cases—industrial gateways, video telematics, and private 5G deployments in manufacturing. For IoT startups, 5G is most relevant for products that require real-time actuation, high-throughput sensor data, or operate in environments where LPWAN coverage is inadequate. However, for most battery-powered IoT devices, NB-IoT and LTE-M remain the more practical cellular options in 2026.

3. What is the Matter standard, and should new IoT products support it?

Matter is an open-source connectivity standard developed by the Connectivity Standards Alliance, backed by Amazon, Apple, Google, and Samsung, designed to make smart home and building IoT devices interoperable across platforms. As of early 2026, over 700 products have achieved Matter certification. Matter 1.5 expanded support to cameras, soil moisture sensors, and energy management systems. For startups building consumer smart home or commercial building products, Matter certification has become a commercial necessity rather than an optional feature—buyers and platform integrators expect it. The certification process has also been streamlined, reducing the development overhead for manufacturers.

4. What cybersecurity regulations apply to IoT products in 2026?

The regulatory landscape for IoT cybersecurity has shifted significantly. In the EU, two regulations are now in play: the Radio Equipment Directive cybersecurity requirements became mandatory in August 2025 for radio-enabled IoT devices, requiring network protection, personal data safeguards, and fraud prevention measures. The EU Cyber Resilience Act has a phased implementation beginning September 2026 (vulnerability reporting) and reaching full compliance requirements in December 2027. In the United States, the FCC's Cyber Trust Mark program is creating a voluntary IoT security labeling scheme.

5. What role does edge computing play in IoT?

Edge computing reduces network strain by processing data closer to the source, lowering latency and improving response times. Over the years, edge computing has matured from an architectural option to a mainstream deployment pattern for industrial IoT, smart cities, and autonomous systems. The edge-AI convergence described in FAQ 1 is accelerating this shift, as more devices can now run intelligent workloads locally. For startups, the practical decision is determining which data needs real-time local processing and which can be sent to the cloud—and designing your architecture to handle both.

6. Can IoT startups benefit from focusing on specific sectors?

Yes, industries such as healthcare, smart cities, and agriculture are ideal for IoT innovation. Healthcare IoT is especially active: CMS broadened reimbursement coverage for connected glucose monitors, blood pressure cuffs, and oximeters in January 2026, enlarging the US remote-monitoring device market by $1.2 billion. Startups that focus on verticals with clear, quantifiable ROI—downtime reduction in manufacturing, energy savings in buildings, yield improvement in agriculture—can build outcome-based monetization models that are easier to sell than generic IoT platforms.

7. What are the challenges startups face in IoT product development?

Startups must navigate device and data security, interoperability across devices and platforms, and solution scalability. In 2026, hardware reliability has emerged as a more prominent failure point than previously acknowledged: a 2025 Eseye study found three in four failed IoT projects were undermined by poor hardware design, with connectivity failures being the leading symptom. This points out the necessity of extensive hardware testing before production and selecting hardware partners with a proven track record in your target environment.

8. What strategies should IoT startups follow to attract investors?

Focusing on robust, scalable solutions that address real-world operational problems remains the strongest investor case. In 2026, investors are particularly attentive to three things: (1) a credible IoT monetization strategy that goes beyond hardware sales—recurring software revenue, data licensing, or outcome-based contracts; (2) edge AI differentiation, as products with on-device intelligence command higher margins and create stronger moats than pure data-collection devices; and (3) regulatory readiness, especially for EU markets, where CRA non-compliance can block market entry entirely. Demonstrating that your IoT product development process incorporates these factors from the start significantly strengthens your funding narrative.

9. How long does custom IoT product development take?

Timeline varies significantly by product complexity, but here are realistic benchmarks for 2026: IoT product discovery (requirements, architecture, vendor selection): 4–8 weeks. Hardware prototyping using off-the-shelf boards: 2–6 weeks. Custom PCB design, testing, and pre-production: 6–18 months. Software ecosystem development (firmware, cloud infrastructure, apps): 3–6 months, running in parallel with hardware. Certification: 2–6 months depending on markets and product category. Total time from discovery to market-ready product: typically 12–24 months for a custom hardware IoT product. Software-only or beacon-based IoT products can move significantly faster. Budget for iteration—most IoT products go through 2–3 hardware revisions before reaching production quality.