Consumer IoT

Predictive Analytics or Predicted Disappointment?

3 min readThe importance of good data for automated, self-learning products The ability to leverage [...]

3 min read

The importance of good data for automated, self-learning products

The ability to leverage data to automate the behavior of IoT devices is one of the most compelling features of the IoT. But to work properly, analytics-driven IoT devices need to be based on useful data. Despite advances in technology, many IoT-enabled devices continue to rely on poor (or siloed) data sources that skew the devices ‘learning’ process. The end result is a market full of shiny products that confuse the user and disrupt the experience.

Christopher Tozzi wrote a compelling article about his experiences with one of the most sought after IoT products, the Nest. While the device is aesthetically pleasing and delivers on its promise for remote monitoring and control, the functional benefits are limited. One of its biggest shortcomings is its inability to accurately learn from and adapt to the surrounding environment. But this problem isn’t limited to Nest. In fact, it’s a recurring problem faced by many device manufacturers.

At CES, the annual electronics show, thousands of vendors occupy the shop floor hoping to penetrate the consumer electronics space. Last year we saw an increased focused on Artificial Intelligence (AI) and self-learning smart homes. But almost a year later, we continue to see product manufacturers unable to properly learn from their device data. In order to enable artificial intelligence, and truly automate the behaviour of IoT devices, manufacturers are tasked with the unique objective of looking at their device data both in aggregate and individually. Where individual data limits the ability to provide a complete, end-to-end, automated experience, aggregate data provides the complete picture required to construct actionable product insights.

When we have aggregate data, we can drill down to see interactions between products and customers, or products and products. Device manufacturers can determine when there is a problem and the relevant context around that problem. This not only allows them to uncover new insights that were not conceived during the initial project planning, but it also proactively reduces the number of product faults, provides feedback back to the product roadmap, and improves aftermarket services.

In his article, Christopher highlighted the devices limited interface as a burden to the user experience. Given that new devices cannot display as much information as their traditional counterparts, product managers must determine beforehand which information to make readily available. But once your devices are deployed, how do you determine whether users find the default information useful or not? One simple way is through the interactions with the device and the application. User interaction with the application is one way to determine if someone is continuously seeking out additional information because it is not available on the device itself. In Christopher’s case he had to open the application to get the current temperature because he wasn’t interested in the projected temperature that was displayed on the device. Overtime, you can learn from the user interactions what information is important to them and adapt the default options.

The availability of data is transforming the IoT. But to progress beyond obtrusive products with limited learning functionalities, device manufacturers will need to focus on the quality (versus the quantity) of their data.

With quality data, IoT analytics can empower product manufacturers with insights on product usage and performance, asset health, customer engagement and much more. They allow you to learn from interactions and determine what users define as ‘comfort’.

Find out how mnubo is tapping into the benefits of product data by transforming raw sensor data into actionable business insights. Read more here!