Can we measure the gap? 

David Lemon
David Lemon
-Jan 06, 2023

Standardization is often seen as a way to improve safety in the workplace. It has been used to drive behavior and work methods in an attempt to reduce errors, mistakes, and at-risk behaviors. However, some experts have argued that the strict adherence to procedures in practice is not always achievable and can lack the flexibility needed to meet operational needs.

The emergence of Safety Differently, Safety II and Human Organizational Performance (HOP) has argued that the adherence to procedure in practice is unattainable, and lacks the flexibility needed to deliver operational needs (Dekker et al., 2014; Wears, 2015).

Let’s begin with defining the gap!

The work set in in detailed procedures is often described as work-as-imagined, or WAI. Work is carried out in practice, work-as-done, or WAD (Hollnagel, 2017). What is basically agreed upon in the literature is that there is a disconnect between the way work is planned and work as done i.e WAI vs WAD. This creates a scenario where there is a difference or gap between set procedures and competing working priorities (Lawrom, 1998; Phipps et al., 2008a).

In the book “Doing Safety Differently,” the authors, Dekker and Conklin, explore the gap between what we expect to happen on the job and what actually happens in reality. When we imagine a task, we often have a specific idea of how it should be done, and we may not consider all of the potential risks involved. Due to the ever changing dynamic nature of the workplace, workers are obliged to improvise new ways while  executing a task. In many cases it works out fine and the task is completed safely. However, on rare occasions this can lead to accidents and injuries. 

To what degree or width of this gap determines the outcome? 

One study (Ashraf et al 2022) followed several people around with cameras and analyzed 524 steps. Key findings were: Most (66%) of the steps were carried out as prescribed by the procedure and in about a third (34%) of the steps, WAD didn’t equal WAI. This included completing a step in a procedure in a different order, partially completing a safety critical step or missing a step entirely.

Given the pivotal importance, how can we measure the gap in a reliable way, or can we? 

With the developments in computer vision in recent years, a novel way begins to emerge to measure and quantify the gap, providing previously unknown insights around WAI vs WAD. As Artificial Intelligence (AI) continues to evolve, existing cameras in facilities can be utilized to analyze workplaces 24/7, generate real-time alerts and data-driven insight to show the scale of the gap. At Intenseye, we can process over 22 billion images daily, generating over 15 million real-time notifications, helping executives make informed decisions to keep over 500,000 employees safe across 25 countries. 

How best to close the identified gap?
To close the gap, the literature recommends a few strategies:

  • By involving workers in the planning process,
  • adopting a just culture1,
  • using risk assessment tools

By using these strategies, we can better understand and address the risks involved in a task. What Intenseye can observe is extremely powerful and could prevent workplace injuries by closing the gap with the help of real-time data and insight. However, the data must be handled with extreme care. Our strategy is to work closely with our customers to ensure that they adopt a human centric approach2 to deploy AI. Our focus is on understanding the gap which makes compliance difficult for the workforce and fixing these systemic issues together with them, rather than blaming the workers. Involving employees and securing their buy-in earlier in the implementation helps to ensure that the alerts generated by AI are used to create a psychologically safe work culture, whereby workers are empowered to speak up in pursuit of finding resolutions. 

A Case Study

One of our customers has a metal shearing area, and no one is allowed within the footprint of the machine while it is operational. There is a control panel within the predefined safe area that is accessed while the shearing machinery is operating. The management is aware of this issue and has a plan to move the control panel. However, what they did not know is that over 300 times in a month employees are putting themselves at risk to meet the job requirements. It is very common that leaders are aware where improvements need to be made, but what is not known is the frequency of a deviated or at-risk behavior. Manual reporting processes are grossly under reporting events due to a whole host of inherent flaws. What Intenseye does is, to provide the true number of at-risk behaviors, i.e. real frequency of deviation. By providing the frequency of events, we are able to help the highest risk become prioritized and the most critical gap between work as done and work as planned gets closed. 

Beyond WAI vs WAD: Any other gap that we can close?

We have observed that intenseye’s real-time data fosters an organic constructive dialogue between decision makers and the frontline staff in order to collectively address the issues at the holistic level. This way, intenseye contributes to closing the gap between the sharp end and the blunt end, culminating in a transparent work culture and better engaged workforce. 
In one of our clients’ EHS directors’ words:

“Intenseye is the voice of the frontline.” 

Can you imagine the impact that you can create through scaling AI to all your facilities across the globe?

I am deeply thankful to Gökhan Yıldız, the esteemed co-author of this piece, for his invaluable contributions! 


1- “Just culture” means creating an environment where workers are not afraid to report near misses and other safety concerns.

2- As an example, Human Performance Principles are the commonly accepted set of principles which endorse the fact that it is not the people who make violations willingly, it is the system or context around them which drives certain behaviors leading to incidents.


S. Dekker, J. Bergström, I. Amer-Wåhlin, et al. Complicated, complex, and compliant: best practice in obstetrics. Cognit. Technol. Work, 15 (2013), pp. 189-195, 10.1007/s10111-011-0211-6

R.L. Wears. Standardisation and its discontents. Cognit. Technol. Work, 17 (2015), pp. 89-94, 10.1007/s10111-014-0299-6

R. Lawton. Not working to rule: understanding procedural violations at work. Saf. Sci., 28 (1998), pp. 77-95, 10.1016/S0925-7535(97)00073-8

D.L. Phipps, D. Parker, E.J.M. Pals, G.H. Meakin, C. Nsoedo, P.C.W. Beatty. Identifying violation-provoking conditions in a healthcare setting. Ergonomics, 51 (2008), pp. 1625-1642, 10.1080/00140130802331617

E. HollnagelCan we ever imagine how work is done? Hindisght, 25 (2017), pp. 10-13

Ashraf, A. M., Peres, S. C., & Sasangohar, F. (2022, September). In Proceedings of the Human Factors and Ergonomics Society Annual Meeting (Vol. 66, No. 1, pp. 1805-1808). Sage CA: Los Angeles, CA: SAGE Publications.

S. Dekker, T. Conklin. Do Safety Differently, Feb 2022, 979-8413008652

#Blog Post
#Journey to Zero
Schedule a Demo