The rapid creation of data from IoT devices
As the cost of sensors, intelligent cameras, and IoT networks continues to fall, the ability for organisations to amass data about people, places, conditions, and assets has never been easier. The proliferation of IoT styled data collection will continue to increase exponentially as organisations pursue their IoT strategies or simply connect more intelligent assets as part of their normal technology and asset renewal cycles.
While the IoT opportunity list can be long, the big value categories are customer experience leading to retention, competitive advantage and sales, and operational effectiveness, especially when designed to optimise ‘systems’ within and external to the organisation.
These broader examples use IoT based monitoring to inform greater levels of asset usage, human interaction, and behaviour, automation of service events as well as identifying opportunities to optimise resource utilisation. This can include assets such as buildings, plant, equipment vehicles, energy, spaces, etc.
The idea of creating substantial business and customer value from IoT data has been written about at length. However, many organisations seem to stall at the data collection step or only use the IoT data to monitor asset use — or at best trigger basic maintenance or operational activities. Why is this?
IoT data must be treated differently from business data
The value of unique IoT data can easily become obscured or lost by storing IoT data in the traditional data lake environment. While data warehousing is a core enterprise facility within many medium-to-large organisations, there is a difference when it comes to IoT data. Accumulating the diverse but often binary or ‘state’ IoT data without understanding the timing and potential source context limits the ability to make sense of the data. At this point, an IoT data ‘dead-end’ occurs.
In the business world, data is often captured in a well-understood state (e.g., number of credit card transactions by customer type, generated as a result of business processes and rules) and is considered as ‘high resolution’ data. By contrast, IoT data can consist of for example electricity meter pulses or pedestrian counts which, as a cumulative data set has a very low resolution or meaning to the untrained eye.
The value of IoT data can become further obscured when traditional BI charting techniques are used to aggregate and normalise data. These data visualisation techniques work well for high-resolution data as the value of the reporting is mainly in identifying trends, volumes, and significant deviations. However, when this approach is applied to IoT data the inherent value of the raw and somewhat binary data is obscured and likely lost forever.
Five critical success factors for using IoT data to generate value
1. Align IoT data collection to business value outcomes
The first critical success factor is to target the collection of IoT data required associated with value cases that come from other case studies or hypothetical considerations. For example, a large scale roll-out of sensors to plant, equipment, supply chain, traffic systems, etc is expensive and produces limited value. An alternative might be to target energy cost reductions for building air-conditioning systems by careful sensor placement to understand service usage times, billing, and building occupancy times.
2. Leverage expertise to validate the local operating conditions
The second is to leverage expertise associated with the specific data source to understand what data will be valuable, how it can be measured, and the likely value that can be created when the data is analysed. An example might be to work with an engineer who understands the detailed engineering considerations for a large building air conditioning plant. The engineer will likely know how the plant functions, what are the normal operating conditions, and what the normal maintenance regime for the equipment is.
3. Specialist knowledge is needed to calibrate the data outputs
The third critical success factor is to leverage the engineering knowledge together with the sensor design and data collection modelling at the source. A specialist IoT engineer can identify the appropriate sensor use case and the optimum installation setup to maximise the depth of data to be captured and the reliability of data collection given the unique environmental conditions. To illustrate this, an IoT engineer knows how to apply vibration measurement to capture data on an audible intensity spectrum as well as frequency.
4. Data science skills are essential to design and automate valued insights
The fourth critical success factor is to design advanced algorithms aligned to the use cases for each IoT data type and their respective value objective. Data science skills combined with algorithm development skills must come together to produce both computational outputs and contextual thinking to identify forecastable patterns, anomalies, and improvement opportunities. When further combined with machine learning capabilities, the ‘training’ of these unique algorithms over time can detect the most acute deviations, leading to more informed business analysis and ultimately inform the ‘next best action’.
5. An “unactioned” insight is a waste
Finally, the fifth critical success factor is business insights, advice, and successfully applying the change. A great insight ‘un-actioned’ is simply a waste. Knowing how to communicate, express business and customer value, and then to apply process/technology/people and data-related changes are critical. This final success factor also involves measuring change success and linking it to continuous improvement.
When analysed and actioned effectively, IoT data can create wide-scale customer and business operational value within and beyond the organisation.
However, given IoT on its own is low resolution, an understanding of how IoT data differs from application data is critical. Five critical success factors enable value to be achieved, ensuring organisations can create significant and worthwhile business value from IoT data.