Embed IoT Functionality in the Product or Cloud?

The Harvard Business Review article, How Smart, Connected Products Are Transforming Competition, identifies ten strategic choices manufacturers must make to capitalize on the Internet of Things.

In this second installment of a series of video interviews, the article’s co-author Jim Heppelmann looks at variables to consider when evaluating how much IoT functionality should be embedded in the product or the cloud.

How much functionality should be embedded in the product and how much in the cloud?

Connectivity serves a dual purpose. It gives us the ability to move information back and forth, but it also gives us a new domain in which to create capabilities. So, for example, as an engineer, given a requirement from the customer I really have four different ways or four different engineering domains in which I can create capabilities.

  • I could use a mechanical approach, and add mechanical features and characteristics.

  • I could use an electrical or electronic approach, and add circuit boards, and those capabilities.

  • I could use software, and put the software in the product as embedded software.

  • Or I could use software in the cloud, and put the software in the data center.

Across the four domains, a product can work pretty seamlessly. So you’re going to find yourself asking the question for a feature or a capability, which of those four domains should I use? And we think that there are some pretty important variables that you ought to think about when evaluating which software should be embedded vs. in the cloud.

  • First of all, what’s the requirement for availability or response time? For example, an anti-lock breaking algorithm inside an automobile probably should be running in the automobile just in case the availability or the response time of the network is a problem.
  • You might also think of the nature of the interface, the human machine interface to the capability. It’s pretty easy to think that rather than struggle to deploy mechanical human machine interfaces, I could easily move the user control of this device up into the cloud, and then deploy it in the rich environment of a smart phone.
  • I’d also give some thought to the nature of the problem we’re trying to solve or the solution that we’re trying to create. If you’re running big data analytics looking for patterns and data coming from many different products, almost by definition you’re going to have to put that capability up in the cloud.
  • And then finally, I think you have to think about the type of connectivity that your thing is going to have to the Internet. This is important because some means of connectivity are essentially free, and others vary with the volume of data being moved back and forth. And I think you want to be careful not to create a big expense around data transmission, moving data up to the cloud if it is actually practical to process the data in algorithms in the product.


Our recommendation is to understand that you have these four domains in which to deliver new capabilities. But then step back and realize that there is an inevitable progression in value and innovation from hardware to software that’s going on. And secondarily realize that within software innovation, there’s an inevitable progression from software that’s embedded in the product up to software that’s running in the cloud. So we encourage you to spend some time thinking about what is the right balance of mechanical, electronic, embedded, and cloud software that’s going to produce the most value in your product and in your business.

Learn More:

In their upcoming October 2015 Harvard Business Review article, How Smart, Connected Products Are Transforming Companies, co-authors PTC CEO, Jim Heppelmann, and Harvard Business School Professor, Michael Porter, will focus on the impact of smart, connected products on companies’ operations and organizational structure. Reserve your copy››