Our Insights

Blogs

Transforming business models with the right IoT data infrastructure

Mark van Leest
Mark van Leest
10 Jun 2021 - 5 min read

Internet oThings (IoT) data allows companies to not only work more efficiently, but it lets them develop completely new business models. In this article, we look at the why and how of transforming from build-and-sell in transport or industry to a data-driven pay-per-use or as-a-service business. 

With innovation storming forward at a dazzling pace, IoT has become such an expansive subject that it has developed very specific subfields. The two most distinct use cases we see are B2C appliance solutions, like smart thermostats, and industrial or logistic solutions, that focus on getting operational data from machines and vehicles. While B2C solutions provide smart, data-based functionality to the user and give companies valuable insights into user behavior, the industrial cases usually aim to make maintenance and operations more efficient. 

Transforming Machine maintenance 

Let’s say you produce machinery that is used for on-site logistics, such as forklifts or airport tractors. Most likely, your product is already equipped with a large array of sensors and telemetry equipment that transmits usage information to your systems. This allows you to tailor youmaintenance schedules, based on detailed usage and climate data, instead of generic measures like mileage or machine age. After all, a tropical climate means the air conditioner is used more often. A cold climate means more wear and tear on mechanical parts. Mileage alone does not tell the whole story either, since short runs produce more wear than long ones. This approach ensures that individual machines receive maintenance when they need it, realizing maximum efficiency while reducing the chance of a machine breaking down. 

Transforming the business model 

This is a great application of IoT data, but it merely optimizes an existing business model. Once you have achieved this level of knowledge of how location, climate and use of their machines impact maintenance and total cost of ownership, you can start looking at your sales and revenue model differently. For example, if your company sells machines or vehicles with a maintenance contract, this usually runs for 3-5 years. After that initial period, your warranties expire, and your clients tend to source their maintenance locally. In order to realize more revenue from this client, you would have to sell them an additional or replacement product. 

Using your data, you could develop a completely new business model. Instead of selling machines, you could sell a ‘pay-per-use' model that charges clients per ride, or an ‘as a service’ model that is paid per month. Using predictive analysis, you can predict when your machines need to be serviced. Shipping an extra vehicle and a mechanical team to the location of the client, you guarantee continuity, and the client can continue business as usual. This method allows you to gain predictable revenue from your products over their entire lifespan. But it is not just beneficial for your company. It also benefits the client. They no longer need to make large investments at once, which lets them scale operations up or down flexibly. Plus, it allows them to outsource the risk of machine breakdowns and guarantee continuity. 

Do you have the infrastructure to support this? 

These applications are not at all limited to logistics. In almost every market, IoT technology is accelerating the shift from selling things for a one-time price to providing solutions as a service or on a pay-per-use basis. This is how cloud data architecture transforms business models and increases predictable revenue for the companies that leverage it. 

What remains, is the question of ‘how’. 

Collecting IoT data from your own machines is probably the easy part. Consolidating, organizing and transforming data to fulfill your analysis needs and combining it with data from other sources like financial systems, ERP and weather data can be a complex challenge. In the end, the measure of success for your data infrastructure is whether validated, usable data flows from your data sources to your front end quickly and effectively enough to enable you to improve your processes, report the KPIs you need to report and grow your business. 

This requires an infrastructure that can create a seamless process, from raw data ingestion to producing actionable KPIs and benchmarks for effective analysis by either your analysts or your machine learning algorithms. But there is no need to start from scratch, as this is a great chance to leverage the full power of the cloud infrastructure you already have. Microsoft Azure, for example, offers the Azure IoT Hub, a fully-managed IoT app platform and an array of machine learning functionalities. AWS is also fully equipped to deal with large amounts of IoT data, with services ranging from device operating systems to fully automated analysis and event detection. 

WHITEPAPER CLOUD DATA STRATEGIES

Building an IoT infrastructure for innovation 

How do you build this infrastructure? How do you get your data initiatives ready for the next level in business model innovation? This is a process with six distinct steps: 

1. Gather more data 

The more data you gather, and the sooner you begin gathering it, the more you can learn from it. Equipment manufacturers need to leave the ‘utilitarian’ paradigm of gathering data or incorporating sensors for the functionality they need at that time. Instead, everyone needs to start ‘thinking in possibilities’Adopt an innovation-oriented mindset and start gathering data because you can, not because you think you will need it. After a while, you will possess large amounts of historical data that allows you to apply machine learning models and see patterns. 

2. Automate ETL

Gathering this much data puts you at risk of getting bogged down with the technical details of extracting, transforming and loading data from many different sources. Gathering data, as well as adding new data sources, should be effortless and easy. Automation is key in achieving this. When building processes for data ingestion, be it continuous, asynchronous, batched, real-time or a combination of these, to storage, preparation, normalization and joining of tables into structured data, aim to keep human intervention to a minimum. Using cloud tools like Azure Data Factory and Azure Integration Services for your IoT data processing makes sure all your ETL processes scale when they need to. Scaling down unused processes keeps costs under control while scaling up when needed prevents bottlenecks in your ETL process. 

3. Build your data warehouse

Your cloud-based data warehouse will be your single source of structured data, optimized for analysis and analytics. There are many options to choose from when it comes to database technologythe most common being Amazon RedShift, Azure SQL and Google BigQueryWhich one you choose is entirely dependent on the data formats you process and your other needs. The important thing here is that you set up your data warehouse as a ‘single source of truth’: any data application should always reference the data in your data warehouse and never depend on one-off integrations or local copiesThis means your data warehouse will have to comply with strict specifications regarding availability and data quality. 

4. Get insights and benchmark


Applying data visualization tools like real-time dashboards, PowerBI, Tableau or Qlick, build dashboards that give you insights into what is happening with your machines or vehicles. Try to find KPIs and benchmarks that work to optimize your current processes and discover new ones. In the machinery example above, you might start by monitoring how often a vehicle is serviced earlier or later than it could or should have been. This will give you a sense of the opportunities for innovation within your current processes and services. 

5. Deploy analytics and AI 

Having sufficient data available and accessible means you can start training machine learning models, using tools like Amazon SageMaker or Azure Machine Learning Studio. These allow you to predict future outcomes of processes and behavior and optimize maintenance or other primary processes. Now, you can move from analyzing and benchmarking historical data to looking ahead, optimizing and gaining competitive advantage. 

6. Develop your new business model


Combining the data insights from your new data collection and analysis infrastructure with your specialist knowledge of your market, you can start experimenting with new business models. The most common way to move forward is to explore outcome-based models like pay-per-use or as-a-service. But there are many other options, such as developing additional services, repurposing unused capacity or providing data to your clients or third parties for their own applications. 

If you have the data, the sky is the limit. 

Do you want to know how we help clients with their data infrastructure? Take a look at our pages on data and analytics and cloud managed services.