Safety analytics - The way to zero harm



Published on December 18, 2018

Safety analytics - The way to zero harm

Zero harm. That is a mantra that any leading organization should have when it comes to safety. The journey to zero harm requires organizations to go far beyond current practices in order to eliminate incidents and accidents in the workplace. What tools and training do organizations need to provide to reduce hazards? Are there certain combinations of workers and supervisors that drive better safety performance? How do we create environments that encourage safe behaviours and build a zero harm culture? The insights needed to answer these questions can be found in the data that exists today.

Good is never good enough for safety

Many organizations possess mature and robust safety programs that rely on industry-leading processes and controls to prevent workplace incidents and injuries. Established practices for reporting, investigations, inspections and audits, equipment and signage, training and hiring, focus on managing known risks and reducing the potential for harm. However, any incident, regardless of severity, leaves a lasting mark on the organization, often leading to extensive downtime and fines, loss of talent and skilled labour, reduced productivity and increased operational costs, and eroded shareholder confidence. As such, the pursuit of continuous improvement in safety is critical for all organizations.

The road to great safety goes through analytics

Often, there is a significant time lag between the occurrence of the incident and the availability of insights into root causes and triggers. Insights that are available lack certainty or detail and provide limited value when it comes to making proactive program changes. Investments in safety programs continue without objective validation of the benefits or the identification of new and emerging risks. Analytics provides a unique advantage in making an impact on the way people make decisions and interact within the organization in order to manage risks that may lead to safety incidents. Incidents are typically viewed in isolation; true insight comes from detecting patterns of repeated behaviours or conditions. A risk-based, data-driven approach to safety management opens up the possibility of uncovering hidden factors and designing specific mitigations.

Exposing hidden risks and driving safe behaviours

Improving safety begins by increasing the awareness of risks, influencing behaviours and attitudes towards safety initiatives, and building an organizational culture that can sustain safe practices. To achieve these goals, safety analytics provides organizations with the ability to identify non-obvious drivers of high-severity accidents and target high-risk operational scenarios and employee groups for interventions before these incidents occur. The key objectives of a safety analytics program include:

  • Generating lagging and leading indicators based on all available data to provide a comprehensive and objective view of current safety performance and key trends;
  • Driving visibility and communications between management, supervisors and employees to enhance the safety culture, and encouraging safe behaviours while retaining top talent;
  • Providing a quantifiable and risk-based approach in order to identify practical safety program improvements and optimize return on investment; and
  • Defining actions and interventions by predicting the attributes and features of future risk scenarios with the ability to generate real-time alerts of unsafe conditions or behaviours.

Figure 1

Figure 1 - Unravelling unknown digital attributes of safety incidents using advanced analytics models

Generating value from descriptive to predictive

Driving the most value from analytics requires a progression from descriptive to prescriptive analytics. Descriptive and diagnostic analytics allow organizations to quantify and evaluate past performance based on available data sources and identify root causes of safety incidents. Predictive and prescriptive analytics build on this foundation of insights and guide organizations towards implementing proactive measures that have the highest potential to reduce safety incidents. Predictive models are not always developed to anticipate a specific event, but rather to quantify the significance of specific variables. Combined, these variables enable managers and supervisors to identify high-risk conditions, develop suitable interventions, and monitor the impact of changes.

Figure 2

Figure 2 – Building safety analytics capabilities from descriptive to prescriptive

As a complement to the analytics concepts described, artificial intelligence (AI) presents an opportunity to expand safety programs beyond traditional structured data sources. Recent advancements in AI have made it possible to analyze text, images and video to identify risks and generate insights that can be integrated into safety practices, investigations and training. AI, in combination with mobile digital capabilities, provides all levels of the organization with the ability to report unsafe practices, hazards and incidents in near real-time, while also gaining access to alerts of dangerous conditions or other safety risks.

Data quality and bias are not barriers

  • Data quality. A lack of data quality, integrity and completeness can certainly erode user confidence and prevent buy-in to act on data-driven insights. Conversely, exploiting existing data, however limited, and providing insights that are valued by users inspires stakeholders to take ownership and improve data collection, including the reporting of near-miss events, which is an invaluable source of leading indicators. Implementing standardized processes, templates and digital tools to enable incident and hazard reporting can facilitate great strides in safety analytics.
  • Bias and preconceptions. Incidents often occur as outliers and a one-size-fits-all approach does not account for these risks. Leading organizations do not see analytics as a replacement for best practices, but rather as an opportunity to leverage insights in order to enhance policies, processes and controls. Communicating insights by simply using visual analytics and interactive dashboards compels managers and supervisors to challenge the status quo and generates momentum to expand analytics efforts. This is the type of culture where analytics can thrive and overcome bias and preconceptions.

Safety analytics key to success

  • Start with the right questions. Begin the analytics journey with a hypothesis or specific areas for discovery (i.e., known unknowns). This direction is needed in order to provide an initial foundation without constraining the analysis and filtering out potentially valuable insights.
  • Think broadly about big data. Leverage non-safety incident-specific sources of data, including operational, financial, HR and training, as well as open source weather and demographic data, to integrate into a safety analytics data set. The emergence of wearables and telematics also provides a wealth of data that can augment traditional safety incident and hazard reports. An enriched data set reveals deeper insights and connections that could not be otherwise detected, thereby exposing hidden risks that can be managed proactively.
  • Make it practical. Analytics and AI rely on subject matter input to validate results and determine what can be implemented, now and in the future. Insights without context are meaningless and will undermine efforts to implement safety analytics.

Analytics brings a new dimension to safety management, empowering decision-makers at all levels to better understand the underlying drivers of performance, proactively manage risks, and realize maximum benefit from safety investments. Influencing behaviours and building a culture of zero harm begins with unlocking the hidden insights in your data.

Safety is not a subject that is commonly associated with the CFO role and, typically, the prevention of accidents and incidents in the workplace is an operational matter. However, the attention to detail and care required to drive strong safety performance are equally relevant to financial performance. Boards and executives of leading organizations see safety as a business differentiator and a key factor in building a culture of excellence across all business units. Strong safety performance helps organizations attract and retain top talent in competitive markets. In contrast, a single safety incident can significantly damage organizational reputation and rapidly erode shareholder confidence. For these reasons and many more, leveraging analytics to improve safety performance is a worthy investment for any leading organization.




Andrew McHardy

Andrew McHardy
Andrew is a senior manager with Omnia AI, Deloitte’s Artificial Intelligence practice in Toronto. He has over 20 years’ experience in engineering, analytics and project management. Andrew works with asset-intensive organizations, developing advanced analytics and AI solutions to improve decision-making and to enhance operational performance related to workplace safety, maintenance, production, asset care and energy management.

Prior to joining Deloitte, Andrew was an engineering officer in the Canadian Army, gaining experience related to project and program management, operations and maintenance, and asset management. He has held leadership roles at the tactical and strategic levels where he was responsible for planning and managing complex military operations, and leading initiatives to enhance organizational readiness and effectiveness.

Andrew is a graduate of the Royal Military College of Canada with a bachelor of engineering (chemical and materials) and a master of defence studies. Andrew also holds a master of science in reliability engineering and a master of science in statistics from Rutgers, The State University of New Jersey.

Correction list for hyphenation

These words serve as exceptions. Once entered, they are only hyphenated at the specified hyphenation points. Each word should be on a separate line.