Talent Matching Notes



Candidate Matching Tool


A tool that allows employers to input the work context and task knowledge they’re looking for in a candidate. The tool reveals occupations that have overlaps with what they’re looking for so that they can search for candidates based on relevant job titles.

This was inspired by AdeptID, who found that Cashiers upskill well into Pharmacy Technicians. The goal was to create a solution to uncover skills and work context overlaps between seemingly unrelated occupations.



Comments

- There are too many questions in the survey, and not all the questions will be relevant to all the occupations. For example, “How often does this job require working in cramped work spaces that requires getting into awkward positions?” is helpful for labour work but not office-based roles.

- The individual filling this out needs to have a good understanding of the role. A lot of the time recruiters/hiring managers responsible for filling the role don’t have as much of an understanding of the role as those perfoming it.


Job to Task Matcher


Occupation overlap for candidates to help them write about their prior work experience in the context of jobs they’re applying for.



I used the Job Task Matcher to make a list for various student jobs and how they may apply to the accounting profession (supplementing the tool with a more detailed version of these overlaps):

Babysitter
- Maintained accurate and confidential records, including observations, activity logs, and medical information.
- Organised and participated in recreational activities, ensuring proper documentation.
- Communicated effectively with parents/guardians about daily activities, behaviours, and developmental concerns.
- Coordinated schedules with other babysitters to manage responsibilities efficiently.
- Performed administrative tasks, such as making phone calls and updating records.

Tutor
- Maintained accurate records of student progress, assessments, and feedback.
- Collaborated with students, parents, and educators to create tutoring plans and share updates.
- Developed lesson plans, study guides, and assessments tailored to student goals.

Barista
- Took customer orders and relayed details to team members for efficient preparation.
- Processed payments and managed cash registers.
- Designed promotional materials, such as signs, to advertise products or events.


Catering Staff
- Recorded food orders and related operational details to ensure accurate preparation.
- Maintained production data, including quantities of special items served.
- Processed customer payments, including calculating totals and handling cash or card transactions.

Valet/Parking Attendant
- Reconciled cash transactions, prepared deposits, and managed cash drawers.
- Notified supervisors of emergencies and contacted appropriate personnel when necessary.
- Assisted clients with directions and vehicle-related needs, such as jump-starting cars.

Hotel Front Desk Clerk
- Maintained guest accounts, including posting charges for rooms, food, and other services.
- Performed bookkeeping activities, such as balancing accounts and conducting nightly audits.
- Communicated with housekeeping and maintenance staff to resolve guest issues.
- Processed payments, computed bills, and handled financial transactions.
- Managed mail and messages, including sorting and distributing incoming communications.

Research Assistant (Social Sciences)
- Collected and recorded research data using tools like surveys, standardised tests, and interviews.
- Conducted data entry and statistical analyses to prepare reports and visualisations.
- Prepared documentation for regulatory compliance, including protocols and presentations.
- Managed project resources, tracked expenses, and supervised team members.

Campus Tour Guide
- Researched and prepared informational materials to guide tours effectively.
- Organised group activities, including planning routes and coordinating schedules.
- Maintained records of attendance and tickets.
- Performed clerical tasks, such as managing correspondence and operating switchboards.

Gardener
- Followed landscaping designs to implement plans, such as planting and sodding.
- Provided consultations on plant care and maintenance, including cost estimates.
- Evaluated designs and work reports to ensure alignment with project goals.Lifeguard
- Observed activities to ensure compliance with safety rules and reported incidents.
- Maintained records of weather conditions, safety issues, and emergency responses.
- Communicated safety concerns to supervisors and requested additional personnel when needed.

Cashier
- Computed and recorded transaction totals, ensuring accuracy in sales records.
- Processed financial transactions, including refunds, exchanges, and credit card payments.
- Reconciled cash drawers and prepared financial summaries for end-of-shift reporting.

Bagger/Packer
- Documented shipment details, such as product dimensions and weights.
- Prepared materials for transport, ensuring compliance with specifications.
- Organised and sorted products for processing or delivery.

Customer Service Representative
- Maintained detailed records of customer interactions, complaints, and resolutions.
- Resolved billing disputes by analysing transactional data and adjusting accounts as needed.
- Recommended improvements to processes and services based on customer feedback.

Receptionist
- Analysed inquiries to provide accurate answers to customers and staff.
- Maintained up-to-date records of staff availability and operational schedules.
- Processed administrative tasks, including proofreading, transcribing, and filing.

Copywriter
- Collaborated with clients to plan and outline projects, ensuring clear goals.
- Verified facts and statistics to maintain accuracy in written materials.
- Prepared and organised content for publication in appropriate formats.

Photographer
- Consulted with clients to determine project goals and resource needs.
- Maintained operational records, including schedules and supply inventories.
- Reviewed and selected the best work from photographic sets.

Visual Content Creator
- Managed archives of digital assets, ensuring accessibility and organisation.
- Conducted market research to align creative outputs with client expectations.
- Designed and prepared files for production and printing.

Competitive Sports Team Member
- Evaluated performance metrics to identify strengths and areas for improvement.
- Represented the team in public events, fostering community engagement.
- Led team initiatives and served as captain to coordinate efforts.

Data Entry
- Verified data accuracy and corrected errors before entry into systems.
- Maintained logs of completed work and ensured compliance with standards.
- Compiled and sorted information for seamless database integration.

IT Support Technician
- Diagnosed and resolved technical issues by consulting manuals and conducting tests.
- Prepared evaluations of hardware/software systems and recommended upgrades.
- Maintained detailed records of transactions, technical problems, and resolutions.





Comments

- The tool requires more detailed O*NET information to describe how the tasks in two occupations overlap. For example, “Execute financial transactions” for Barists relates to “Receive and process customer payments” while in the accounting/auditing profession, this is to “pay charges, fees, or taxes.” A candidate might want to highlight their understanding of these overlaps, and their broader understanding of the role they’re applying for by writing something like “As a barista, I received and processed customer payments, which has prepared me for the skills required to pay charges, fees, or taxes as an accountant/auditor.” 


- This isn’t a very good solution to skills-based hiring. The onus is still on the candidate to identify the overlaps between their work and what employers are looking for. But it’s a start that equips candidates with a better understanding of how to express and contextualise their experiences for what employers might be looking for. 


- It doesn’t work very well. The occupational data from O*NET is not applied semantically. Therefore, this approach wouldn’t work with other taxonomies that are more nuanced that O*NET (for example, ifATE contains statements that have the same meaning but are phrased differently  e.g. “Prioritises health and safety” vs “puts health & safety first”.)


Rate My Credential


It can sometimes be useful (and fun) to build things that won’t work. Rate My Credential is one of those tools.

Users anonymously rate the people they’ve worked with which assigns those ratings to those individuals’ credentials. This can be helpful for deciding who to hire based on qualities certain credentials might reveal.



Comments (why it doesn’t work):
- Misaligned incentives. The data is provided by individuals who don’t benefit from the effort it takes to provide information.

- Bias. Although ratings are anonymous, users are more likely to rate colleagues they’re friends with or colleagues they dislike, believing this will affect the individual concerned. These personal sentiments screw results. This can probably be solved with more clarity and trust about how data is used.

You can play with it here.


Personal Reflections


The majority of employers decide who to hire based on qualifications. Yet many of them don’t fully understand the knowledge that specific qualifications impart. Ask most employers what someone with a CS degree from a top university can do better than someone who has completed a coding bootcamp and I can almost guarantee you that they won’t be able to tell you. That’s because knowledge of computer science isn’t necessarily what they’re looking for or they’d be more bullish about understanding the details. 

Degrees are used to determine IQ and identify the quality of conscientiousness. It’s assumed that IQ is directly correlated to the standardised test scores that grant individuals admission into college in the first place, and that conscientiousness acts as the driving force through the finish line to graduation. Brian Caplan discusses the economics behind this: a student who completes seven of the eight semesters earns half the salary that a student who completed all eight. If the acquisition of knowledge was what employers valued, the number of semesters should directly correlate to earnings. Either that, or half the knowledge is delivered in the final semester (which isn’t the case).

There’s nothing wrong with wanting to hire high IQ, conscientious people. The issue arises when employers aren’t explicitly aware that this is what they’re filtering for when they use credentials to identify talent. An explicit awareness of this might help them manage their disappointment at graduates’ lack of preperation and relevant skills for assuming workplace responsibilities.


There are over 1 million credentials available today, and that number is growing. Yet employers don’t have the tools to understand what knowledge they impart.
What are the broader implications of evaluating candidates based on skills vs credentials?

Inequity compounds: Employment is based on employment history which is based on college degree(s) which are based on admission to college which is based on standardised test scores which are based on access to financial and informational resources (tutors, counselors, parents who went to college, neighbours who went to college).

Skills-shortage = not enough people know how to do the things we need people to do in order to sustain our standard of living (“the economy?”) Some ascertain that this shortage is (partly) based on perception: that we have people who know how to do the things we need people to do but that we lack the mechanisms to identify these individuals. We probably don’t have any secret electricians, but there are shortages in other sectors that alternative methods of talent identification could alleviate to close the skills gap. Even if the packaged talent doesn’t exist, a standardised mechanism for communicating skills would enable the identification of individuals with some of the skills, therefore identifying the most frictionless candidates to possibly upskill into roles.

An ageing population is magnifying the skills-shortage. This is (will) become particularly apparent in tech. Not just because there are more tech companies, but because non-tech companies now also depend on tech to maintain a competitive edge.

i/ The rapid pace at which tech is developing means that traditional education can’t keep up with delivering knowledge that’s relevant and aligned with the needs of the labour market. The only way (right now) to maintain current knowledge of relevant skills is to be employed.
ii/ We won’t have enough young people to fill these roles.

Learned that it’s anticipated that half the world's tech talent will come from Africa by 2050. 

Thoughts about skills, especially how to quantify them:

A taxonomy is a classification system (usually used in biology to classify organisms) that uses a list of hierarchically orgnanised defined terms. A skills taxonomy is just that, but for skills, and can be used to connect skills with occupations. There are a few well-known ones like O*NET, ESCO, Holland Codes, SOC, and the Institute for Apprenticeships and Technical Education (ifATE), and some companies, like Goldman Sachs, even have their own. 

You can combine taxonomies using something called a “crosswalk.” A crosswalk lets different taxonomies “talk” to each other by translating terms between them. For example, one taxonomy might list “Accountant,” while another lists “Financial Auditor.” A crosswalk connects the two so they’re treated as related. This example is a bit simplified and real crosswalks deal with much messier differences in how things are defined.

The idea is to take credentials (like degrees or certifications) and translate them into the skills they actually represent, using O*NET as the foundation. Then, you’d create crosswalks to connect O*NET to other taxonomies. This could help employers figure out what skills a credential actually provides, match candidates to jobs based on those skills, and identify the best ways for people to upskill and fill internal skills gaps. Right now, employers mostly look at the name of a credential (like “BS in Computer Science”) because they don’t have the tools to see the skills behind them, so they make decisions based on assumptions about these skills. If we could add a layer of skills under the credentials, people could be evaluated based on what they can actually do, not just what their degree or certificate is called or where it’s from.

The model would work like a calculator but for words. It takes words, like job titles or skills, and turns them into numbers so they can be analysed objectively. For example, NLP models create embeddings, which are basically numerical representations of words based on their meaning and context. The tricky part is making sure the model works for all possible phrases and contexts, and that it gives consistent, reliable results no matter what input it gets. A good model is like a good calculator: when you put in 1+1 it always gives you 2. Identifying mistakes is also a bit of a pain because language itself can be quite subjective and the definitions of words can vary across different contexts.

For jobs with high risk or responsibility (like a surgeon or an air traffic controller,) the model needs to be stricter about matching a candidate's skills exactly. For jobs with lower stakes (or roles with heavy managerial oversight), it can be more flexible. A formula for this could loosely be based on something like: stakes = risk level + level of supervision. For example, O*NET surveys might show that Supply Chain Managers report error consequences as “not serious at all” or “serious” (35% each). That means the stakes are moderate, so the model wouldn’t need to be as strict as it would be for an occupation like “Surgeon”. 

Here’s a simple example of how skill matching might work. Let’s say a candidate has the skill “apply” (Bloom’s Taxonomy level 3), and a job requires “analyse” (Bloom’s Taxonomy level 4) in the context of “computer science principles.” The context match is 100% because they’re both about the same field, but the level match is only 75% because the candidate’s level (3) is slightly below the required level (4). This means they’d need to improve by 25% to fully meet the requirement. This kind of detailed matching helps employers see exactly where candidates stand and what additional training they might need. If we reverse the example where the job requires a level 3 and the candidate's level is a 4, the match would be 100%.

Bloom’s Taxonomy, which organizes learning objectives into levels like “apply” and “analyse”, is great for understanding theory, but jobs often involve translating theory into practice. That’s why we should add a layer to account for real-world/physical application. This makes it more relevant for workplace skills which go beyond what’s covered in traditional academic settings.

Most recruitment tools right now still rely on keyword matching, basically just pulling terms from resumes and job descriptions. Some tools are more advanced, like Eightfold.ai or SeekOut, and focus on skills-based matching, but these aren’t widely used yet and they still rely heavily on candidate input. That means there’s still a huge opportunity to shift the focus from credentials to skills and build something smarter than what’s out there.

One of the reasons to think all this through now is because once a model is built, adding new variables down the road can be a huge pain. It could mean having to rethink the whole system (annoying and time-consuming.) That’s why we’re considering as much as possible upfront, like whether someone learned a skill in a classroom with mandatory attendance (which implicitly demonstrates an ability to show up/time management) or through a self-directed course (which can illustrate initiative/self-motivation). 

In summary, a taxonomy is like a numbering system for organising skills. A model can use taxonomies to analyse skills objectively by turning words/phrases into numbers and performing calculations on those numbers. The overarching goal is to help employers focus on what people can actually do, instead of just looking at titles of credentials/previous roles. By combining taxonomies like O*NET and Bloom’s Taxonomy, and factoring in things like risk, stakes, and learning environments, we can create something much smarter and more useful than what’s currently available. It’s like building a calculator for skills: you input job requirements and candidate profiles, and it calculates how well they match. The aim is a tool capable of more objective, accurate and equitable candidate evaluation.


Resources


  • Digital Credentials and Talent Acquisition Tech: Closing the Data Gap Between Learning and Hiring - Northeastern University Center for the Future of Higher Education and Talent Strategy [link]
  • Review of Skills Taxonomies - Frontier Economics (Danail Popov, Sarah Snelson, and Thomas Baily) [link]
  • Charting the Future of Assessments - ETS [link]
  • Skills England: Driving Growth and Widening Opportunities - Department of Education (UK) [link]
  • Skills-Based Hiring: The Long Road from Pronouncements to Practice - The Burning Glass Institute (Matt Sigelman and Alex Martin) and Harvard Business School Managing the Future of Work (Joseph Fuller) [link]
  • The O*NET Content Model: Strengths and Limitations - Michael J. Handel [link]
  • O*NET ESCO Technical Report - European Commission [link]
  • Hidden Workers: Untapped Talent - Harvard Business School Managing the Future of Work (Joseph Fuller, Manjari Raman, Eva Sage-Gavin, and Kristen Hines) [link]
  • Apprenticeship + Degree - Paul Fain [link]
  • Small Towns, Big Opportunities - Georgetown University Center on Education and the Workforce [link]
  • NHS Long Term Workforce Plan - NHS [link]
  • Opening the Black Box of Credit Transfer to Everyone - Ithaka S+R (Pooja Patel and Martin Kurzweil) [link]
  • Providing Credit Transfer Visibility to Improve Credit Mobility - Ithaka S+R (Betsy Mueller, Emily Tichenor, Martin Kurzweil, and Alexandra W. Logue) [link]
  • Designing and Implementing Work-Based Learning: A Call to Action for CHROs - Northeastern University Center for the Future of Higher Education and Talent Strategy [link]
  • Skills Needs in Selected Occupations Over the Next 5-10 Years - RAND Europe and The Institute for Employment Research (IER) and the University of Warwick (Joanna Hofman, Dr Julia Doyle, Asha Haider, and Dr Michaela Bruckmayer) [link]
  • Standardized Talent Asset Mapping Protocol (STAMP): Unleashing the Power of Talent Data - Gobekli [link]
  • This outside-the-box approach to embedding certification for in-demand tech careers is paying off outside the high school classroom Laura Aka for Working Nation [link]
  • Globalization, Government Popularity, and the Great Skill Divide - IZA Institute of Labor Economics (Cevat G. Aksoy, Sergei Guriev and Daniel S. Treisman) [link]
  • Working Futures 2017-2027: Long-run labour market and skills projections for the UK - Institute for Employment Research, University of Warwick (Rob Wilson) and Cambridge Econometrics (Mike May-Gillings, Shyamoli Patel and Ha Bui) [link]
  • Putting Nigeria to Work: A Strategy for Employment and Growth - The World Bank [link]
  • Strategic Priorities: Shaping the Workforce and HR Agenda in 2024 and Beyond - The Future of Work Hub by Lewis Silkin [link]
  • The Employment Situation – July 2024 - Bureau of Labor Statistics [link]
  • The Future of Jobs Report - World Economic Forum [link]
  • Understanding the Emerging SkillsTech Landscape - Northeastern University Center for the Future of Higher Education and Talent Strategy [link]
  • What do you want to be? Youth aspirations in the time of the COVID-19 crisis. Evidence from three Sub-Saharan countries - World Bank (Valentina Costa, Ivette Contreras-Gonzalez and Amparo Palacios-Lopez) [link]
  • 2024 Retention Report - Work Institute [link]
  • Career development in Israel: Characteristics, services and challenges - Israel National Employment Service (Benny A. Benjamin), Hebrew University of Jerusalem (Itamar Gati) and The Academic College of Tel Aviv-Yaffo (Hedva Braunstein-Bercovitz) [link]
  • Stackable Credential Pipelines and Equity for Low-Income Individuals: Evidence from Colorado and Ohio - Lindsay Daugherty, Peter Riley Bahr, Peter Nguyen, Jennifer May-Trifiletti, Rooney Columbus, and Jonah Kushner [link]
  • Study: Most Job Seekers Abandon Online Job Applications - Society for Human Resource Management (Dave Zielinski) [link]
  • Counting U.S. Postsecondary and Secondary Credentials - Credential Engine [link] 
  • Experience You: Phase 1 Demonstration Report - U.S. Chamber of Commerce Foundation, Education Design Lab and T3 Innovation Network [link]
  • Opportunity Era of the Higher Education for in the Talent Economy - Pearson [link]