What is an IoT Platform? (And How to Choose One)

What is an IoT Platform? (And How to Choose One)
Daniel Elizalde
What is an IoT platform

If you’re confused about IoT platforms, believe me, you are not alone. Many Product Leaders have the same challenge. It’s a complex topic and one that I get a lot of questions about.

In this post, I’ll give you a clear picture of how to approach IoT platforms, including:

  1. What is an IoT platform and why you need one
  2. How IoT platforms fit into your overall product strategy
  3. The different categories of IoT platforms across the IoT technology stack
  4. Key considerations when choosing your IoT platform
  5. How to decide between building or buying your IoT platform

Let’s get started!

What is an IoT platform, anyway?

Think of an Internet of Things platform as a group of technologies that provide the building blocks for developing your product. IoT platforms provide the “infrastructure” you use to create the specific features of your solution.

The goal of an IoT platform is to provide all the generic functionality for your application so you can focus on building features that differentiate your product and add value for your customers.

By taking over the non-differentiated functionality, IoT platforms help you reduce your development risk and cost, and accelerate your product’s time to market.

When people talk about IoT platforms, they often launch into technical jargon like transport protocols, rules engines, data lakes, etc. While those considerations are important and deserve thoughtful planning, they don’t clearly illustrate how an IoT platform can help you.

Let’s break down the key tasks that an IoT product needs to perform, highlighting the functions your IoT platform should cover. An IoT product needs to:

  1. Acquire real-world data via sensors
  2. Analyze data locally (edge computing)
  3. Connect to the cloud to transmit data and receive commands
  4. Store data in the cloud
  5. Analyze data in the cloud to create insights
  6. Command the “things” to perform specific tasks based on insights
  7. Present insights to users

Additionally, there are important “behind-the-scenes” capabilities IoT platforms should provide:

  1. Perform all operations securely across the IoT technology stack
  2. Identify and manage all of your IoT devices (at scale)

Based on this simplified description, a good IoT platform should provide the tools and infrastructure to cover as many of these tasks as possible.

For example, if your Internet of Things platform is very strong on analytics but doesn’t help you transport data from devices to the cloud, then you are left with a big gap. In practice, you are likely to use more than one IoT platform in your product, but more on that later in this post.

How IoT Platforms Fit with Your Product Strategy

Before launching into product development and IoT platform selection, you need to make sure you have a solid IoT product strategy.

A robust strategy needs to cover these three areas:

  • Desirability (does the market want your product?)
  • Viability (is there a business model where you can make money with your product?)
  • Feasibility (can you build your product?)

IoT platforms help you with feasibility by accelerating and reducing risk in your development process.

But keep in mind that feasibility should only be considered AFTER you have desirability and viability. In other words, you need to make sure you are solving a real customer pain and that your solution will lead to financial gains, before building your product.

Since this post is about IoT platforms, I’ll focus on the Technology Decision Area of my IoT Decision Framework.

IoT Decision Framework - Technology

This post assumes that you already completed the UX, Data, and Business Decision Areas, meaning you’re ready to select your IoT platform.

Make sure you start with the UX Decision Area to gauge Desirability and then complete the Data and Business Decision Areas for Viability.

If you are not familiar with the IoT Decision Framework, check it out! The rest of this post builds on those concepts.

Recommended article: A Product Management Framework for the Internet of Things

Understand the Various Categories of IoT Platforms

IoT products are very complex because they have to integrate multiple components across the 5 layers of the IoT technology stack.

IoT Technology Stack

Given this complexity, it is not likely that you’ll find one single IoT platform that covers all areas of the IoT technology stack. Therefore, you will need a few different types of IoT platforms to cover the whole spectrum.

The most common categories of IoT platforms are:

  • Cloud platforms (a.k.a. application enablement platforms)
  • IoT connectivity platforms
  • IoT device platforms
  • Analytics platforms

IoT Cloud Platforms (a.k.a. Application Enablement IoT Platforms)

This category of IoT platforms provides the core building blocks for your product, including consuming, transporting, storing, analyzing, and displaying data. As the name implies, their aim is to enable the rapid development of your application by abstracting the complexities of building an IoT solution.

There are hundreds of Internet of Things platforms in the market so it can be daunting to figure out which one to use. Application enablement platforms come in all flavors, including:

  • Industrial platforms
  • Consumer platforms
  • Platforms targeting developers
  • Higher-level (drag-and-drop) platforms, good for prototyping or MVPs
  • Platforms focusing on specific verticals
  • On-premise vs. Edge vs. Cloud platforms

There is ample information available on the wide variety of IoT platforms: Gartner has started tracking IoT platforms in their Magic quadrant, and IoT Analytics has several great reports on this topic.

For context, some of the top players in the IoT Cloud platform space include:

  • Microsoft Azure IoT
  • AWS IoT
  • Hitachi Vantara
  • PTC

If you’d like an inside look at Microsoft’s Azure IoT, I recommend these two episodes of my IoT Product Leadership podcast where I interview Sam George, head of Azure IoT.

IoT Connectivity Platforms

Connectivity is an integral part of the IoT technology stack, linking on-the-ground IoT devices to the Cloud or any other big data repository.

Many IoT products (mostly in the consumer space) rely on Wi-Fi to connect to the Internet. For these products, connectivity management might be included in their application enablement platform. Therefore, spending the extra time and effort to guarantee connectivity may not be worth the investment.

On the other hand, products that are part of what’s called “critical IoT,” such as manufacturing equipment, connected cars, or the smart grid require more robust communication technologies to reach the Internet. For example, many of these IoT devices rely on cellular technologies like 4G, 5G, or NB-IoT.

In this scenario, the data doesn’t flow directly from the IoT device to the internet and into the Cloud platform. Instead, the data travels through a cellular network—managed by a mobile network operator—before being routed to the internet.

There are many benefits of leveraging third-party infrastructure to transfer data from the IoT device to the Cloud, including dedicated bandwidth, strict connectivity SLAs, and added security.

But working with network operators introduces new challenges, including added complexity when connecting the IoT device to a mobile network, and maintaining partnerships with cellular carriers all over the world to ensure your product can operate globally.

This is where IoT connectivity platforms come into play.

Their role is to seamlessly manage the connectivity between your devices and your Cloud platform, across many cellular operators (or other types of specialized network operators such as SigFox).

Instead of you contracting with multiple carriers, connectivity platforms provide a single interface for deploying, monitoring, and managing all your devices around the world. These platforms often include additional functionality such as traffic monitoring, connectivity management, geolocation, device management, over-the-air updates, and device provisioning.

Examples of IoT connectivity platforms include Curiosity by Sprint, Jasper by Cisco, IoT Accelerator by Ericsson, and Pelion by ARM.

As the industry develops, the line between Cloud and connectivity IoT platforms is starting to blur. Cloud players such as Microsoft and Amazon are implementing capabilities included in IoT connectivity platforms. Likewise, companies building IoT connectivity platforms are looking to go “upstream” and build the functionality previously only provided in Cloud IoT platforms.

IoT Device Platforms

IoT device platforms provide hardware building blocks for developing IoT devices. And considering that hardware is usually the “long pole in the tent” for IoT product development, it is important to consider what already exists in the market to accelerate your efforts.

If you’d like to dive deeper into the components of an IoT device, I recommend this article: What is an IoT Device?

Selecting the right IoT device platform depends on where you are in the IoT product adoption curve.

The chart below shows my recommended strategy for building IoT devices.

IoT device hardware through product lifecycle

Your goal is to align your IoT roadmap with the green blocks. Very early on, during the discovery stage, you want to leverage as much off-the-shelf hardware as possible. Only when you approach product-market fit and start driving towards deployments at scale, is it time to invest in custom hardware.

The reason is that early on, you don’t have validation of what exactly solves your customer’s problems. Therefore, it’s better to spend your time and money running quick experiments, as opposed to building expensive hardware that you don’t know your customer wants.

As you gain traction in the market, your focus will shift from product-market fit, into the pursuit of profitability. That is the time to invest in custom hardware for miniaturization, reduced cost of the bill of materials (BOM), longer battery life, better supply chain agreements, improved serviceability, etc.

So what does an IoT device platform look like? Here are some examples for each stage of the IoT adoption lifecycle.

IoT device platforms – Pre-product-market fit

Your goal at this stage is innovation. This requires you to understand your customer’s needs and propose potential solutions to address them. Your focus should be to create quick prototypes you can share with your potential customers and iterate as quickly as possible. This is the perfect time to leverage full off-the-shelf hardware components to build your prototypes.

You can use single-board computers with sensors such as Arduino or Raspberry Pi. If you need industrial hardware, you can try hardware platforms like PXI or Compact RIO from National Instruments. You can also use a laptop and attach sensors via USB or a PCI card. Or use a smartphone or tablet since they already have a lot of sensors.

For an example use case, listen to this episode of my IoT Product Leadership podcast where I interview author and executive coach Barry O’Reilly on how he used iPhones to prototype an IoT solution for a transit system.

IoT device platforms – When approaching product-market fit

As you approach product-market fit, you might need to develop increasingly customized IoT devices that more closely resemble the functionality and form-factor of your final product.

For this stage, you can also leverage hardware components like Arduino, Beagle Bone, Raspberry Pi, or the OEM version of CompactRIO as the core of your device. Your hardware engineering team can use these building blocks as part of your own custom hardware to create a semi-custom IoT device.

At this stage, your focus is not only functionality but also form-factor. The IoT device enclosure, hardware user interface, etc., are elements that require the focus of both your engineering and industrial design teams.

IoT device platforms – At scale

Once you find product-market fit and are ready to scale, it’s time to invest in custom hardware. Your goal is to optimize for form-factor, cost, supply chain, and serviceability.

The IoT device platforms you used pre-market fit might be too expensive or bloated to deploy at scale. At this stage, it’s common to completely redesign your IoT device to meet your specific product and company goals.

The IoT device platforms you’ll encounter at this stage, usually come directly from chip manufacturers such as Intel, ARM, Nordic, and Xilinx.

You can also work with either Cloud vendors or connectivity vendors who can help you select and integrate the best hardware for your IoT product. They can also assist you with reference architectures and partners that can help with your IoT device design and manufacturing at scale. For example, Verizon offers this through ThingSpace.

As I mentioned above, the lines between the various platforms are starting to blur. Today, you’ll find chip manufacturers like ARM, who have robust connectivity and Cloud platforms. Or Cloud providers like Microsoft who have a robust IoT device platform (Azure Sphere).

You can learn all about Microsoft’s Azure Sphere platform in this episode of my IoT Product Leadership podcast.

Note: Keep in mind that an IoT product should integrate seamlessly from the IoT device, all the way to the Cloud. To do so, you’ll need to run specialized software on your IoT device. This software is often provided by the IoT Cloud platform companies in the form of an SDK (software development kit). Make sure that the hardware platform you select is compatible with your Cloud platform.

IoT Analytics Platforms

The goal of an IoT product is not to collect data. It is to provide actionable insights to your users. That’s why analyzing the data is as important (if not more) than collecting the data.

Most of the Cloud platforms already include analysis tools, which may be enough for many applications.

But if your application has additional requirements around visualization, data processing, digital twins, artificial intelligence (A.I), or machine learning (ML), an IoT analytics platform can accelerate your IoT development.

Some of these platforms offer generic analytics capabilities that you can customize, while others offer specialized, vertical-specific capabilities such as logistics, asset tracking, or predictive maintenance. Examples of IoT analytics platforms include Watson by IBM, C3 AI, SparkCognition, and UpTake.

To dive deeper into this topic, I recommend these episodes of my IoT Product Leadership podcast: Deep dive into Azure Digital Twins

How to select the right IoT platform?

Choosing the right IoT platform(s) can be daunting. There are hundreds of options and vendors at each layer of the IoT Technology Stack. You’ll need to do your research to determine the best option for you.

If you are not familiar with the IoT space, it’s a good idea to get a third-party perspective from industry analysts like Gartner or IoT Analytics. Then you can zero in on specific vendors that fit your needs.

When selecting an IoT platform, you need to consider much more than just the technical capabilities of their solution. You are looking for a combination of technical, business, and operational capabilities that align with your company strategy and where you are in your product adoption lifecycle.

For example, if you are early in your journey, you should focus on IoT platforms that can help you build and test prototypes very quickly. Scalability, cost, and feature set should be less of a concern.

As you move towards market fit and scale, the focus should shift towards IoT platforms with more scalability, stability, and a global footprint.

Keep in mind that when you are transitioning to scale, you might need to completely re-platform your product, and that’s okay. As Product Leaders, we need to set clear expectations with Executives and Investors. As your product adoption increases, it shouldn’t be a surprise to them that your team will need to spend time refactoring the product for scale, which may mean transitioning to a new set of IoT platforms to support this new stage.

Here are 5 key areas to look for when selecting an IoT platform:

  1. Reputable company. IoT is risky enough as is. Trusting the core of your product to an unknown company might backfire. Make sure you evaluate their reputation, stability, financials, and track record.
  2. Large ecosystem. IoT is so big that there’s no way a single company can dominate it all. Looking for a company with a strong app and partner ecosystem will be a good investment in optionality and expansion. Most of the top IoT platform providers don’t do hardware themselves, but have a strong partner ecosystem to pull from. That is always a good sign.
  3. Open APIs. Extensibility will be key, so make sure you select a provider that gives you programmatic access to as much of their functionality as possible.
  4. Vertical focus. Aligning with a vendor that understands your industry is always a plus. Their solution will be designed to handle your type of data, analytics, and even help you comply with industry regulations.
  5. Strong onboarding. Adopting a new platform is not trivial. Look for companies that have a strong solutions department (or professional services) that can train your team, help you with the architecture, and guide you through the proof of concept stage.

Ultimately, you are looking to build a partnership. It’s a complex decision, and one you can’t make alone. It’ll require you to work closely with various groups in your company, including Engineering, UX, Data Science, Finance, and more.

Should you build or buy your IoT platform?

By now, it should be clear that there are many benefits to leveraging commercially available IoT platforms. But believe it or not, many companies, particularly those who are engineering-driven, believe that they need to build every single piece of their IoT solution.

These companies spend years and millions of dollars building non-value-added infrastructure, instead of focusing on building differentiated features to serve their customers.

So let me be clear. You simply don’t need to build the complete IoT infrastructure yourself. There’s no point reinventing the wheel.

By using commercially available IoT platforms, you reduce your development costs because you have more functionality sooner, with less engineering effort. This means you can focus your engineering team on what really matters: your core value proposition.

Think about it. IoT platform vendors have large teams of developers improving features, fixing bugs, and making sure their offering is rock solid. That is their business, so it makes sense for them to invest in it. By leveraging their work, the quality and stability of your product will ride the wave of their investment.

Let’s debunk some of the most common myths I hear against using commercially available IoT platforms.

Myth #1: My product is unique

IoT platforms are meant to be generic, so most likely you’ll find some gaps between their feature set and your ideal solution. The key here is to evaluate the size of that gap and determine if you can get around it. Are the missing features a must-have? Are there other platforms or software vendors who can cover the gap?

Also, ask yourself if that functionality can wait. One of the big advantages of jumping onto an IoT platform is that you piggyback on their growth. Top companies have hundreds of developers working on enhancements and new functionality. The gap you are experiencing might be covered in the next release, not to mention the new functionality, partners, and tools they’ll continue to provide throughout the years.

Focus your time on developing the uniqueness of your product. That’s where your money will be best spent.

Myth #2: IoT platforms are expensive

Yes, platforms cost money. But in the long run, these expenses are small compared to what it’ll cost you to build your own. Building a working IoT platform (software or hardware) will take years and millions of dollars. On top of all that development cost, you need to add support and maintenance costs. Forever.

Instead of creating an asset, you’ve probably created a liability. In a nutshell:

[tweetherder]You CAN’T build an IoT platform for less than what it would cost to adopt a commercial one.[/tweetherder]

Even if you are able to pull off the enormous technical challenge, in the end, you’ll have something that’s not core to your strategy, is not differentiated, and is probably 2–5 years behind where the industry is today. As a Product Manager, it would be difficult to justify that investment.

Rich Mironov debunks the “do-it-yourself illusion” in this great post.

Myth #3: Investors will give my company a lower valuation

Many Product Leaders are worried that investors will see less value in their product if they are leveraging third-party components or IoT platforms. Nothing is farther from the truth.

Investors care about unique value propositions and ROI. How would you explain to them that you spent their money reinventing the wheel? Not a conversation I look forward to.

Let me put it this way: saying that your product loses value if you build it on a platform is like saying Tesla cars have no value because they don’t make their own tires. Or saying Netflix has no value because they build it on top of AWS. I don’t think so.

The Bottom Line

Building an IoT product is very complex. By outsourcing non-core areas of your product to an IoT platform, you’ll reduce risk, minimize cost, improve quality, and accelerate your time to market.

Oh, and one last thought. If you are not thinking of using an IoT platform, I assure you your competitor is. Don’t be surprised if they beat you to market and are able to provide more value-added features faster. Consider yourself warned.

17 Comments

  1. Hari Narayana 1 year ago

    Thanks for the informative content.

  2. Mike Hadlee 3 years ago

    Thanks for a great article Daniel. This outlines lot of points that need to be considered. However, many Industrial IoT project have not see the light past PoC’s . There are a multitude of reasons and the primary one being technology overdose. Example: IoT Platform providers providing public cloud over Azure, AWS, GCP etc. This is just making it a cost plus plus for the end user and a longer payback or benefit curve.

    We have tested and using Foghub (https://foghub.io) which directly shares clean contextual machine data to the destination IT system of choice of the end user. For us this has proven to be a very lean approach where the data from our manufacturing assets directly lands in the best of the breed IT system of our choice example: ERP, BI, Dashboards, Analytics, BPM and messaging applications.

  3. Melody Kopp 3 years ago

    Thanks for sharing it, Daniel.

    Many people use IoT devices in their lives but not many people aware this much deep on IoT Platform. You have shared it in-depth. Keep sharing.

  4. Danny 4 years ago

    Nice article, do you think there will be different criteria for selecting platform for consumer iot vs iiot . I mean some of platforms are more suitable for one over other

    • Simith Nambiar 2 years ago

      Although there are difference fundamentally the things you would look out for in IIoT Platform use cases are the data ingestion rates, cost, filtering mechanisms provided at the Edge, Machine learning inference at the edge, Time series data storage and analysis, Stream Analytics and Software to run at the edge provided by the vendor and it goes without saying all of this having a strong focus on security. For consumer IoT, you would need to gain an understanding of User identity Management, How do you handle Security of Devices (X.509 certificates) , Certificate Authorities, Device Management – launching new features – Over-the-air (OTA) updates, Analytics to drive product and customer insights Remote monitoring and Analysis etc., Serverless technologies you can use – As you want to drive down the cost of maintaining the Consumer product over it’s life time with connectivity, telemetry and OTA.

  5. dileep k 5 years ago

    The concerns regarding adopting an IoT platform are rightly negated with having the right questions to ask about the platform – which you mentioned. In addition, an IoT product manager can focus on security aspects of what is on his plate (device hardware and device software).

    Do you have a write up on what an IoT platform comprises? Any short article? Do you cover them in your course?

    Your analogy about Tesla not making its own tires drives home the point. Nothing like a good analogy.

    • Daniel Elizalde 5 years ago

      Hi Dileep,
      Thanks for your comment. In my IoT PM certificate program, I do cover what comprises an IoT platform as well as the criteria to drive the vendor selection and build vs buy processes. Check it out at courses.danielelizalde.com.

      -Daniel

  6. Tom Raftery 5 years ago

    I’m curious why you left the SAP’s Leonardo platform out of your list of top platforms – it ticks all the boxes you mention (reputable company, large ecosystem, open platform, vertical focus, and strong on boarding).

    In full disclosure, I work for SAP, but even if I didn’t, I’d still be asking

    • Daniel Elizalde 5 years ago

      Hi Tom,
      Thanks for your comment. I know SAP is doing great work. There are over 450 platforms out there, so it’s hard to include them all. I’ve received similar emails from various platform providers, so I might update the post in the future.

      What would you say are SAP’s Leonardo strengths compared to some of the platforms I mentioned here?

      -Daniel

      • Tom Raftery 5 years ago

        Thanks Daniel,

        SAP is a company which has been creating digital twins of business processes for 45 years now. Our Leonardo platform is designed specifically to take the information from the Things, and to enhance an organisation’s processes with the insights gained from them.

  7. Arun Sinha 5 years ago

    Great Post. Right on point.

    • Daniel Elizalde 5 years ago

      Thank you Arun!

  8. Marc Phillips 5 years ago

    This post is perfect. Buy infrastructure (it’s battle-tested by experts who make it great for a living), and build intelligence (where the value to your business is and where your team should spend all their time). An additional key point is that ‘platform’ is becoming less useful a term than ‘solution’. For example, complex solutions (optimizing farmed fish production, autonomous heavy equipment operation, distributed medical devices, etc) often integrate multiple ‘platforms’, such as AWS and Azure for secure data collection, storage, and advanced analytics/machine learning, with an IoT data management ‘platform’ like Bright Wolf Strandz (http://brightwolf.com/platform/) to address data normalization and transformation, granular access control, and device lifecycle management issues.

    • Daniel Elizalde 5 years ago

      Hi Marc,
      Thanks for your kind words and for adding more insights to this conversation.

      -Daniel

  9. Antonio Sánchez 6 years ago

    Thank you very much, your blog have been useful and really clear. Looking forward for more information, please contact me if you could come to Mexico and share with us your knowledge and experience. I’ll be in contact.

    • Daniel Elizalde 6 years ago

      Thank you!

  10. ROBERTON 6 years ago

    Pocket IOT is the best Iot platform : http://www.ouroboros.fr/Pocket_iot_solution.html

Leave a reply

Your email address will not be published.