There’s an interesting development going on in the embedded-systems industry. Initially, data was only used for internal purposes and quality assurance. Customers would send log files to product companies who would analyze them to figure out why the product wasn’t operating as it should and what to do about it. Over time, the periodic data sets have turned into more or less continuous data streams and the data collected has evolved from being concerned with QA to focusing on product performance and measuring value delivery to customers.
As the volume and expenses associated with collecting and storing data have increased, companies have been investigating ways to create novel value from this data through direct or indirect monetization. We can identify at least four phases that companies go through.
The first step is where the company gives the data away as part of the overall product offering. Typically, the data is processed and provides nice dashboards for customers to gain an understanding of the product’s performance. However, as the customer gets this for free, there’s limited focus on the data part of the total offering. This is similar to how, in many industries, software was given away for free as part of the mechanical or electronic product. We’re now getting paid for software, but many are now giving data away for free.
The second step is where the company has developed some form of data-driven service to customers using the data from each specific customer. Here, the first monetization of the data starts and even if it often is a minor revenue stream, both customers and the company itself are now, in fact, benefiting from the collected data.
Once the second step is in place, often customers ask the company how they perform when compared to others. This is where the third step is initiated as it allows the company to provide data-driven services to customers using data from all customers. Now, customers can benchmark themselves and understand where to improve and where to extend their lead over competitors.
The fourth step is where the company moves to find alternative markets/customers for the data from its primary customer base. Here we see the start of a two-sided market where the primary customer base generates the data that is then monetized with a secondary customer base. If played right, this can allow the company to transition from a product to a platform company and to ignite a thriving business ecosystem where the company can ‘tax’ transactions between ecosystem partners and thus create highly profitable revenue streams that, in time, may outweigh the revenues from products.
In our discussions with companies in Software Center, there are three main challenges that companies struggle with, ie pricing, disruption risk from suppliers and partnering. The first challenge, pricing, is simply concerned with putting an actual value on data sets or data streams. The preferred model, though difficult to execute on, is value-based pricing, meaning that you estimate the value that the receiver of the data gets from it and then negotiate a fair share of that value.
The second challenge is that product companies are constantly asked by suppliers for data. Initially, this concerns data from the subsystem provided by the supplier, but over time, it tends to broaden and cover a larger and larger scope. The risk becomes that, with enough data, suppliers can become powerful competitors in data-driven services. They often serve multiple companies in the same industry and if they manage to negotiate data from all of them, they’re much better placed to generate a competitive advantage. Of course, many companies have little interest in this, but finding the right balance between sharing and avoiding creating a new competitor is a difficult one. The best practice seems to be the insertion of a control point, meaning that you can cut off a supplier at any point in time when it becomes clear that they’re starting to compete with you.
Finally, even for potential partners from other industries that are interested in gaining access to the data collected by the company, it’s often very difficult to decide which of these potential partners are worthwhile to participate with and which ones should be ignored. There are few generic guidelines here, but in general, a potential partner that can help you build a two-sided market and, in due time, become a platform company is much more valuable than alternatives.
The embedded-systems (or cyber-physical-systems) industry is becoming increasingly aware of the importance of data but is struggling with operationalizing this awareness into a solid business. I’ve outlined the typical pattern that I see companies follow, as well as the key challenges experienced. Engaging in data is very difficult for companies that still think of themselves as metal-bending experts, but it’s critical to get going. Not using your data, or just giving it to someone else to build a business around, is the worst thing you can do. For all the risks and challenges, in a digitalizing world, you need to be world class at software, data and AI and the only way to achieve that is to experiment and learn. Go digital!