The Evolution of Hiring Part 1. The AQYR team looks at how recruitment is becoming more intelligent, evidence-led and human where it matters.
Most talent teams open a new requisition the same way: write the advertisement, post it, brief the agencies, start again from zero.
Meanwhile, the person who can do the job may already be in the system. They applied fourteen months ago for an adjacent role, reached a second interview, and lost to someone with slightly more site experience. Nobody has looked at their record since.
This is not a failure of effort. It is a limitation of architecture. The applicant tracking system was built to record and manage applications, not answer the question “who do we already know who could do this?”. As hiring becomes more data-rich, that distinction matters more.
The cost lands hardest where the market is thinnest. Close to three in ten occupations assessed by Jobs and Skills Australia were in national shortage in 2025, and well over a hundred have been in shortage every year since 2021, concentrated in exactly the trades and technical groups that resources and infrastructure employers draw from.1 At the same time, Australian job advertisement volumes are falling while applications per advertisement sit at the highest level on record.2 Candidate databases are filling faster than at any point in the sector’s history, and almost nobody is going back into them.
The ATS is a system of record, not a system of discovery
Applicant tracking systems were built to solve a compliance and workflow problem: prove that a fair process was followed, move applicants through stages, produce an audit trail. They are extremely good at this, which is why effectively every large employer runs one and why more than nine in ten use it to filter or rank candidates at the first screen.3
Discovery is a different problem. It asks the system to reason across every person it has ever seen, infer capability from incomplete evidence, and produce a shortlist for a role that did not exist when those people applied. That needs a data model organised around people and what they can do. Most are organised around requisitions and who applied to them.
So a candidate does not really exist in the system as a person. They exist as a set of applications. Search the database and you are not searching a talent pool. You are searching a filing cabinet indexed by job opening, and by the end of the last decade the average corporate opening was filing roughly two hundred and fifty of them.4
Six things keep good people invisible in there.
Keyword search makes capability easy to miss. Most database search is still literal string matching with filters bolted on, so it works only when a candidate used the employer’s vocabulary. In technical hiring, equivalent work is described in different ways: shutdown scheduler and turnaround planner, fixed plant and processing plant, HV switching and high voltage authorisation. Employers know the consequence: nearly nine in ten surveyed executives agreed that qualified high-skilled candidates were screened out because they did not match the exact criteria in the job description, rising further for middle-skilled roles.5 Requirements accumulate until the filter stops behaving like a sieve and starts behaving like a fine mesh.6
The record is a snapshot, and it decays from the moment it is created. Just over a million Australians changed employer or business in the year to February, with Mining and Construction both above the national rate.8 Credentials move as well. A Western Australian high risk work licence runs for five years, can be renewed up to twelve months after it expires, and past that window the holder has to retrain and be reassessed before doing licensed work again.9 Medicals lapse, rotations finish, tickets are added, roster tolerance changes, contract preference flips to permanent. None of it reaches the system, because the candidate has no reason to log in and nothing prompts them to. The upshot is that the reason someone was screened out usually expires before the record does. The person rejected eighteen months ago for a lapsed licence may have renewed it the following week. The system still says unsuitable, permanently, for a reason that stopped being true almost immediately.
Rejections are coded, not explained. Dispositioning captures an outcome, not progressed, unsuccessful at interview, position filled, and discards the reasoning. The reasoning is the valuable part, and it is usually specific to the role rather than the person. Someone rejected as too senior is a strong candidate for the next role up. Someone who would not take a 2:2 is viable the moment an 8:6 opens. Someone rejected because a better candidate existed is, by definition, a good candidate. Runners up are the most expensive version of this loss: people already assessed, already worth hours of somebody’s time, already known to be competent, and a straightforward route to a fast hire when a comparable role reopens.10 They are filed under a label that means no, with no memory of why.
The same person exists five times. Records fragment fast. Someone applies directly in 2023, is submitted by an agency in 2024 under a different name format, comes back through a referral, and appears again in a recruiter’s LinkedIn project that never touched the ATS. Add mergers, separate instances across business units, legacy systems kept after a migration, agency submissions living in email, and spreadsheets built for one project. Without identity resolution the system has no idea these are one person, so each fragment holds part of the picture and none holds all of it. Even a perfect search returns a misleading record.
Nobody is measured on database utilisation. Time to fill rewards fresh, warm applicants; the internal database produces colder candidates whose data may be wrong and who need re-engaging. That work looks slower even when it produces better, cheaper hires. The Hidden Workers authors make the broader point: talent acquisition metrics tend to reward expense minimisation rather than the quality of what is acquired, and would be better aimed at time to full productivity, attrition and advancement.11 Almost no organisation tracks what share of hires came from people already in the system, so the work stays invisible in reporting and behaviour.
Retention rules and consent quietly shrink the usable pool. Holding candidate data is neither free nor unconditional. Australian Privacy Principle 11.2 requires reasonable steps to destroy or de-identify personal information once it is no longer needed for any purpose for which it may be used or disclosed.12 The employee records exemption does not rescue anyone here: it covers current and former employment relationships and expressly does not extend to prospective employees who are not subsequently hired, which is precisely the population sitting in the database.13 Most teams respond by retaining everything and hoping, or purging on a schedule and losing the institutional memory. Neither is a strategy.
Why the shift matters
Agency fees on roles that could have been filled internally are the easy number for a business case. The bigger opportunity is the value already sitting inside the organisation.
That opportunity is avoiding searches that have effectively already been run. It is recognising the same person across multiple applications, reusing assessment work already completed, and strengthening the only talent pool an employer owns outright: people who have already put their hand up.
“Every operator I speak to is worried about where the next crew comes from,” says Andrew Hornby, Managing Director of AQYR. “Almost none of them can tell me how many people are sitting in their own system who have already applied, already interviewed, already said yes to the site and the roster.
That is the cheapest and warmest pool any of them will ever have. It is quietly rotting, because nothing in the stack was ever built to keep it alive.”
What the next model looks like
This is not a case for replacing the ATS. It is a case for adding the intelligence layer the ATS was never designed to provide.
Resolve identity first. Deduplicate and merge across every source holding candidate data, so one person is one record. Nothing else works reliably until this does.
Index capability, not keywords. The system should understand that a shutdown scheduler and a turnaround planner are the same job, that a Certificate III in Engineering (Mechanical Trade) implies a set of competencies, and that eight years on a fixed plant transfers to a comparable site. That is a structural relationship between skills, roles, sites and credentials, not a synonym list.
Filter affirmatively rather than negatively. Most systems winnow by excluding people for what they lack. Configuring instead for the skills and experience that actually predict performance in the role is both more efficient and more inclusive.14
Track the things that expire. Tickets, licences, verifications of competency, medicals and clearances all have known validity periods. A system holding an expiry date can flag currency, rather than deferring to a screening decision made eighteen months ago.
Capture why, not just what. Structured rejection reasons, roster and location preferences, salary expectations and mobility, recorded at the point of decision, while the information is free to collect and somebody still knows the answer.
Build a refresh loop. Periodic, genuinely low effort prompts for candidates to confirm what has changed. It doubles as consent renewal and turns a decaying archive into a maintained asset.
Measure whether any of it is being used. Track three numbers: the share of hires already in the system before the role opened; final-stage runners-up later placed elsewhere; and records contacted or updated in the last twelve months. The third is usually the confronting one.
Search the inside before the outside. Make internal rediscovery the first step on every requisition, and make it fast enough to use.
That matters because a recruiter’s week is already consumed by looking. Part two is about where those hours go.
The candidates are already there. The next stage of hiring is making them easier to rediscover, understand and engage.
Bibliography
Australian Bureau of Statistics. Job Mobility Continues to Decline into 2026. Canberra: ABS, July 2026. https://www.abs.gov.au/media-centre/media-releases/job-mobility-continues-decline-2026
Fuller, Joseph B., Manjari Raman, Eva Sage-Gavin and Kristen Hines. Hidden Workers: Untapped Talent. Boston: Harvard Business School Project on Managing the Future of Work and Accenture, September 2021. https://www.hbs.edu/managing-the-future-of-work/Documents/research/hiddenworkers09032021.pdf
Government of Western Australia. Renew a High Risk Work Licence. Perth: WorkSafe Western Australia. https://www.wa.gov.au/government/multi-step-guides/high-risk-work-licensing/renew-high-risk-work-licence
Jobs and Skills Australia. 2025 Occupation Shortage List, Key Findings Report. Canberra: Jobs and Skills Australia, October 2025. https://www.jobsandskills.gov.au/sites/default/files/2025-10/2025%20OSL%20Key%20Findings%20Report_0.pdf
Office of the Australian Information Commissioner. Australian Privacy Principles Guidelines, chapter 11. Sydney: OAIC. https://www.oaic.gov.au/privacy/australian-privacy-principles/australian-privacy-principles-guidelines/chapter-11-app-11-security-of-personal-information
Office of the Australian Information Commissioner. Employee Records Exemption. Sydney: OAIC. https://www.oaic.gov.au/privacy/privacy-guidance-for-organisations-and-government-agencies/organisations/employee-records-exemption
SEEK. SEEK Employment Report, July 2026. Melbourne: SEEK Limited, August 2026.
Society for Human Resource Management. How to Rediscover and Re-Engage Top Candidates. Alexandria: SHRM, November 2024. https://www.shrm.org/topics-tools/news/talent-acquisition/how-to-rediscover-and-re-engage-top-candidates
Hiring is evolving is a series examining how recruitment is changing, and what better systems can make possible. Part two looks at where a recruiter’s week actually goes, and how more of it can be returned to judgement and relationships.
Jobs and Skills Australia, 2025 Occupation Shortage List, Key Findings Report (Canberra: Jobs and Skills Australia, October 2025). Two hundred and ninety three of one thousand and twenty two assessed occupations were in national shortage, twenty nine per cent, down from thirty three per cent in 2024. One hundred and thirty nine occupations have appeared on the list every year since 2021, concentrated in Technicians and Trades Workers. ↩
SEEK, SEEK Employment Report, July 2026 (Melbourne: SEEK Limited, August 2026). National job advertisements fell six per cent year on year, while applications per advertisement rose to their highest recorded level. ↩
Joseph B. Fuller, Manjari Raman, Eva Sage-Gavin and Kristen Hines, Hidden Workers: Untapped Talent (Boston: Harvard Business School Project on Managing the Future of Work and Accenture, September 2021), 3 and 20. Ninety nine per cent of Fortune 500 companies run an applicant tracking system. More than ninety per cent of surveyed employers use their recruitment system to filter or rank candidates at initial screening, ninety four per cent for middle skilled roles and ninety two per cent for high skilled. ↩
Fuller et al., Hidden Workers, 12. ↩
Fuller et al., Hidden Workers, 3. Eighty eight per cent for high skilled candidates, rising to ninety four per cent for middle skilled. ↩
Fuller et al., Hidden Workers, 22. Seventy eight per cent of the business leaders interviewed estimated that filtering eliminated half or more of their middle skilled candidates, and eighty per cent said the same of high skilled candidates. ↩
Fuller et al., Hidden Workers, 15. The employer survey covered 2,275 executives and the worker survey 8,720 respondents, across the United States, the United Kingdom and Germany rather than Australia. The software stack and the practice are substantially the same here. ↩
Australian Bureau of Statistics, Job Mobility Continues to Decline into 2026 (Canberra: ABS, July 2026). Just over one million employed people, seven point two per cent, changed employer or business in the year to February 2026. Mining and Construction both recorded nine per cent, against a national rate of seven point two. ↩
Government of Western Australia, Renew a High Risk Work Licence (Perth: WorkSafe Western Australia). Licences run for five years. Renewal is available up to twelve months after expiry, though licensed work cannot be performed in that gap, and beyond twelve months the holder must retrain and be reassessed. ↩
Society for Human Resource Management, How to Rediscover and Re-Engage Top Candidates (Alexandria: SHRM, November 2024). ↩
Fuller et al., Hidden Workers, 4. ↩
Office of the Australian Information Commissioner, Australian Privacy Principles Guidelines, chapter 11 (Sydney: OAIC), paragraph 11.3. The obligation does not apply where the information forms part of a Commonwealth record or where an Australian law or court order requires retention. ↩
Office of the Australian Information Commissioner, Employee Records Exemption (Sydney: OAIC). Once an employment relationship is formed, pre-employment records held about that individual do become exempt. ↩
Fuller et al., Hidden Workers, 4. The report frames this as a shift from negative to affirmative filter logic. ↩
AQYR is a workforce intelligence system built from industry experience, market intelligence and AI, powered by millions of workforce, salary and labour market data points held in a proprietary knowledge graph. AQYR assesses every candidate and explains why they fit, integrates with the rest of the recruitment process, and uses AQYR Insights to map a role before it is advertised. AQYR operates as a hybrid of software, recruitment services and fully managed delivery. Learn more at https://aqyr.ai/.