Job Resilience FAQs

The Job Resilience chart enables you to inform ongoing business continuity and workforce planning to enable productive remote work opportunities beyond the COVID-19 pandemic.

Also in this article:


If you are looking for data to inform the return to workplace strategy, the Job Resilience chart will support your evidence-based approach.

The Job Resilience chart will help you to identify how 'remoteable' each job is – meaning how flexible it is to work from home, as well as the degree of human interaction is necessary to perform the job (greater risk of contagion).


What does the Job Resilience module help me to understand?

Use this feature to understand:

  1. The degree to which each job in your workforce can be performed productively while remote based on underlying tasks, enabling interventions for select job cohorts.
  2. The degree of human interaction in different jobs within your workforce and how this, when coupled with remote work potential, can inform workforce agility and resilience planning.
  3. The work-related exposure of high-risk workforce cohorts based on age and location, to enable mitigating strategies with their safety (given pandemic or other disruptive events) in mind.

How is the degree of remote productivity defined?

Degree of remote productivity refers to the percentage of total tasks within a job that can be completed effectively while remote via enabling technologies.

For example, a Biomedical Engineer’s task to 'install, adjust, maintain, repair, or provide technical support for biomedical equipment' requires in situ work within a lab or medical setting. On the other hand, a Biomedical Engineer can continue to complete other tasks remotely, such as 'adapt or design computer hardware or software for medical science uses'.

A task that can be performed remotely via enabling technologies is, therefore, a remoteable task and contributes to a job’s overall degree, by percentage (%), of remote productivity. The model is agnostic to the task and does not consider whether an organisation would like this task to be performed remotely or not, only if it can be.

For example, the majority of face-to-face training could be replaced by online training and videos. Similarly, all interviews and many doctors' appointments that don't require physical examination could also be replaced by a video call. These types of tasks would be classified as remoteable.

How is the degree of human interactivity defined?

Degree of human interactivity refers to the percentage of total tasks within a job that require interaction with another person. For example, a nurse’s task, 'administering medication to or monitoring patients for reactions or side effects' requires human interaction. This contributes to the overall degree, by percentage (%), of human interactivity in this job. Similarly, an information security analyst’s task, 'conferring with users to discuss issues such as computer data access, security violations, and programming changes' also requires human interaction.

Another example is a bus driver’s task to 'comply with traffic regulations to operate vehicles in a safe and courteous manner' is also interactive. While there is nothing in the task description that indicates it is interactive, the fact that it is a bus driver who is undertaking this task in a bus full of people makes this task interactive.

How is job remote productivity calculated?

Faethm has built a neural network to predict the level of remote productivity and human interactivity of each job. This is predicted at the task level and then aggregated to the job level based on the task-time allocated to each task.

The neural network is trained on 1,550 labelled tasks and applied to the entire task database of more than 20,000 tasks. Tasks are classified as remote productive or not if the nature of the task can be completed away from its core workplace. Technology that enables remote productivity can support remote tasks. These must currently exist or could likely be implemented within a three to a six-month timeframe.

How is job human interactivity calculated?

Faethm has built a neural network to predict the level of remote productivity and human interactivity of each job. This is predicted at the task level and then aggregated to the job level based on the task-time allocated to each task.

The neural network is trained on 1,550 labelled tasks and applied to the entire task database of more than 20,000 tasks. Tasks are classified as human interactive if the task requires any level of human interaction, whether it's face-to-face or a call. It also includes people working in the public domain whose work environment means they are in close contact with the public, such as a bus driver or waiter.

What is the level of prediction accuracy?

The model is highly accurate. Remote productivity prediction has a precision of 82.4% and recall of 87.2%, while human interactivity has a precision of 89.5% and recall of 83.4%.

Precision and recall are defined as:

  • Precision: the ratio of correctly predicted positive observations to the total predicted positive observations.
  • Recall: the ratio of correctly predicted positive observations to all real observations in the class.

For example, if we had built a model to predict whether a fruit was an apple or not in a basket of fruit, then, precision is the number of times an apple was correctly predicted divided by the total number of times an apple was predicted. Recall is the number of times an apple was correctly predicted by the total number of apples. 

How is the geographic location of the workforce taken into account?

The remoteability of a task is impacted by the ability to physically work from home and is dependent on a number of factors such as access to the internet. Economic indicators are applied to determine the potential to remote tasks for all countries.

Example indicators include available infrastructure; percentage of individuals with home internet. These indicators are applied to the task remote productivity classification to adjust a job’s overall remote productivity. Data is derived from the World Economic Forum: The Global Competitiveness Index and Innovation datasets 2007-2020.

How are the different job impact groups and zones defined?

The quadrant zones serve as an indication of the level of impact that workers can experience in their jobs as the pandemic becomes more prevalent.

A higher degree of human interaction and lower degree of remote productivity increase the level of job impact independently, as employees who are unable to work remotely and have a higher level of human interaction may be most exposed to contracting COVID-19. The limits of each quadrant have been defined so that each one contains a similar percentage of unique jobs. This was defined using Australian Census data. 

How are the different job impact groups and zones calculated?

The zone boundaries were set in such a way to split the jobs across the four key categories. The goal of the boundaries is to group similar jobs together so that appropriate change action can be used for each category. There is a relatively even distribution of jobs across categories, but with a slightly lower proportion in high risk compared to low risk to enable a more targeted response for the higher-risk roles.

How is the interactivity threshold modelled?

The interactivity threshold is set to 0.6. This means jobs with an interaction score >= 0.6 are high interactive jobs while those with an interactivity score < 0.6 have low interactivity levels. Most jobs have a base level of interactivity. This can be face-to-face, via phone or email, or just serving the public in a public space. All these interactions are included in the interactivity score. Accounting for a higher base level of human interactivity drove a requirement to increase the threshold to classify highly interactive jobs from regular levels of interaction.

How is the remoteability threshold modelled?

The remoteability threshold is set to 0.4. This means jobs with a remoteability score >= to 0.4 have high remoteability while jobs with a remoteability score <0.4 have low remoteability. While most jobs will have some tasks that are tied to a human location, not all these tasks will be essential for BAU. As a result, only jobs with less than 40% of their roles classified as non- remoteable are considered in the highly non-remoteable category.

What filters can I apply to this feature?

The Job Remoteability feature is responsive to descriptive filters including org unit, location, age, gender, level, and any custom filters enabled. The feature and modelling are for a current state view of the workforce only, and therefore future scenario filters such as year and technologies cannot be applied. Leveraging the descriptive filters, particularly age and location, can help identify key impacted or vulnerable cohorts within your organisation.

Are emerging technologies taken into account?

No emerging technologies are taken into account to model the degree (%) of job remoteability or degree (%) of job human interaction. The Job Remoteability feature only assumes access to current technologies such as teleconferencing, chat systems, 5G connectivity. The feature, therefore, shows what jobs can be enabled, via investment in current technologies, to work or interact remotely.

How can I download the results / raw data?

To download the insights visible in the chart and table (both interconnected), select the download icon at the top right-hand corner of the page. This will provide the options to download the data as either an Excel or CSV file. The export will be reflective of the data on screen and any applied filters. Users can customise the download by applying the filters that reflect the scope of their analysis.

How does Faethm avoid demographic bias in data modelling?

Faethm believes in practicing the ethical application of data. No race, diversity, or gender information is applied in Faethm modelling that would impact client decision making.

In the Job Corridor, for example, we do not utilise gender or age information when calculating our job fit scores. This means we have removed biases related to age and gender when presenting possible transition pathways from current to future jobs. There are minor current known biases in Faethm data and models. Faethm modelling of jobs, skills, and tasks are derived from US seed data (O*NET). Assessment of automation and adoption is heavily weighted to Western developed countries. We are actively working to diversify our data sources to solve for this.

How do headcount and FTE differ?

FTE refers to Full-Time Equivalents, or the sum total of jobs that are impacted, irrespective of the number of individuals in the jobs. For example, two half time (0.5 FTE) employees will be displayed as 1 FTE. Headcount refers to the actual individuals that reside in the jobs, and more accurately reflects the number of people who will need to be supported through any transition. Both FTE and Headcount figures are provided in the Business Resilience Module to provide continuity with the rest of the platform, while also providing additional detail on the people impacted.