An in depth guide to how the Job Corridor model works
Job Corridor models a broad range of data points to calculate how closely-matched one job is to all other jobs in the Faethm Occupation Ontology. Based on this calculation it generates recommendations of the best matched jobs.
It then provides detailed information on the attribute gaps between two jobs, based on the user's selections. The model works out the projected size of each gap using Faethm's five-level proficiency scale.
Note on terminology and concepts
This article uses a range of terms that you may either, be seeing for the first time, or that may have a unique meaning in the context of the Faethm platform. Selected terms are hyperlinked to full definitions in a dedicated Job Corridor Terminology and Concepts article
Job Corridor models two types of career transitions, each with different data inputs:
Why the difference?
When using the "Assess Career Pathways" transition a user selects a current job first. This makes the current job the fixed, reference job i.e. the one job that all potential recommended jobs must be compared to. Assess Career Pathways is generally going to guide more proactive transitions to help workers move away from jobs that are likely to be automated in the next 10-15 years. It's therefore important that jobs with weaker future demand and low resilience to technological change are filtered out.
When using the "Fill Future Jobs" transitions a user selects target job first, making the target job the fixed, reference job. An organisation’s demand for the target job may be relatively urgent, so it’s less important that the target job has strong long-term demand or low automation risk. The organisation’s priority is filling the role quickly and/or meeting more targeted and certain demand for the job.
Job attributes are the 244 fundamental elements that Faethm uses to map and define all the jobs in the Faethm Occupation Ontology. A job is defined using attributes by assessing the level of required proficiency for each attribute according to Faethm's five-level proficiency scale. Attributes create a standardised, universal set of characteristics that can be used by data models, like the one behind Job Corridor, to perform detailed comparisons of different jobs at scale.
There are 244 job attributes broken into 6 categories:
When the Job Corridor model compares jobs using attributes, it looks at their relative required proficiency levels.
Job Skills are also used to define and map jobs but unlike attributes, skills are unique to each job. There are some skills that are common to a lot of jobs, like communication skills for example, but there are very few jobs that are made up of the exact same skills.
Faethm maintains a comprehensive list of standardised skills that are used to define all jobs called the Faethm Skills Ontology. As all jobs are defined using this standardised skills ontology, the Job Corridor model can assess how closely the skill requirements for different jobs align.
Jobs can require large numbers of skills, many of which are common to a lot of jobs. So when the Job Corridor model compares jobs using skills, it compares the top 20 most relevant skills for each each job and assesses how many top-20 skills the jobs have in common. In assessing how closely two jobs align, the model also takes into account how high the skill is ranked for each job and whether skills that might not be matched are still closely related.For example, the reference job may require Tableau as a skill and even though another job may not have this skill, it may have a related skill like powerBI. While matched skills are indicators of closer alignment, related skills are still relevant. All else equal, it will be quicker and easier to transition to a target job from a job with unmatched but related skills, than a job without related skills.
The job fit assessment - a simplified three-step description
Job fit assessments have been done for all jobs in the Faethm Occupation Ontology. This process summary describes how the job fit assessment works for a single reference job as this is how the user experiences the model. The scores described below are all calculated by the model but not shown to the user. They are discussed here for the purpose of explanation only.
1. Job attributes and skills comparison - match score
The Job Corridor model compares the required skills and attributes for a reference job to the required skills and attributes for all jobs. It generates a match score from 0-100 for all jobs. The closer the score is to 100, the more skills and attributes that a job has in common with the reference job.
Remember that all jobs are defined using the same attributes, so the attribute comparison is about comparing the gap in required proficiency for each attribute. The more attributes for which a job has the same or lower proficiency requirements than the reference job, the higher the match score.
2. Job transition comparison - weighted score
Not all jobs are have the same level of transferability. For example, highly specialised jobs like nursing are not made up of skills that are common to a lot of other jobs. For these specialised jobs even the best fit will have relatively low match score. This could be interpreted as these jobs not having any suitable transition options.
It also means that other jobs, like customer support manager, have a lot of skills in common with other jobs. For these jobs a lot of jobs will have a relatively high match score. This may make it difficult to differentiate between the larger number of recommended jobs with high scores.
To address the varying level of transferability between a specialised and more common jobs, the Job Corridor model does a second calculation that adjusts all of a reference job's match scores according to how they compare to the match scores of all other jobs. This means that jobs with less-transferable skills will have their match scores adjusted up and jobs with highly-transferable skills will have their match scores adjusted down.
This adjusted score is known as the weighted score and it is the weighted score that is used to classify recommended jobs into the job fit tiers that a user sees.
The weighted score is still on a 0-100 scale. The closer the score of a recommended job is to 100, the better the fit to the specific reference job.
3. Unsuitable roles filtered out
When using the "Assess Career Pathways" transition, three filters are applied before classifying jobs into job fit tiers and showing them to the user as recommended jobs. These filters rely on assessments and projection from data models outside of the Job Corridor model.
- Job level - only roles that are the same seniority level as the first-selected role or 1 level above or below will be shown to the user. All other roles are removed from the potential pool of recommended jobs.
- Demand for job - jobs that are in lowest quartile of demand, measured in number of job ads per year, are removed from the potential pool of recommended jobs. If the first-selected job is in the lowest quartile of demand, then this filter is not applied.
- Automation risk - jobs that are projected to have an automation rate of >15% over the next 15 years are removed from the potential pool of recommended jobs. If the first-selected job has an automation rate of >15%, the threshold is adjusted up in proportion to the automation rate of the first-selected job
When using the "Fill Future Jobs" transition, only one filter is applied:
- Job Level
Classifying recommended jobs into job fit tiers
There are 5 job fit tiers. Jobs that are most similar to the reference / first-selected job in terms of required skills and attribute proficiencies are sorted into higher tiers i.e. closer to top tier fit.
Recommended jobs are sorted into tiers based on their weighted job fit scores (0-100) (see above). The threshold scores for each tier are based on the overall distribution of job fit scores based on the top 20 transitions for each job (approx. 5000 transitions). As there are over 5 million potential transitions in total between all jobs, and the Job Corridor model analyses all of them, the model ends up finding more transitions in lower job fit tiers than higher ones. If the thresholds weren't set as specified below, then the over-representation in lower job-fit tiers would skew the data and represent potentially suitable transitions as undesirable.
The job fit tier is a relative measure. This means that the job fit for all potential recommended jobs is unique to each reference job. Or put another way, the 2nd tier recommended jobs for reference job A, may have a different transition gap than the 2nd tier recommended jobs for reference job B.
Regardless of the transition gap, the highest tier recommended jobs for a given reference job are the best fits for that job.
How the transition analysis is generated
Once a user has selected a current and target job, the transition analysis provides some of the detail that went into the job fit assessment.
Job attribute gaps
The gap between the the required proficiency for the current and target job is the core part of job fit assessment. It's visualised on a scale where each of the five proficiency levels represents an increase in overall proficiency in an attribute of 20%. The transition analysis groups the attribute gaps into four categories:
- large - the proficiency requirement for the target job is more than a full proficiency level above the current job i.e. at at least a 25% gap
- medium - the proficiency requirement for the target job between a half and a full proficiency level above the current job i.e. a 10-25% gap
- small - the proficiency requirement for the target job is less than half a full proficiency level above the current job i.e. less than 10%
- no gap - the proficiency requirement for the target job is the same or lower than the current job i.e. 0 or negative gap
Top Job Skills
Top skills for the target job are shown. They are drawn from the Faethm Skills Ontology, which is regularly updated based on the latest job ad data.