Skills-Based Hiring Has Entered Phase Two

The Requisition That No Longer Started With a Job Ad

At 9:11 on a Monday, a talent leader at a large industrial company opened a role that would have looked familiar in 2021 and strangely outdated in 2026.

The business needed someone who could manage a maintenance planning workflow, work across plant operations, understand a new layer of AI-assisted scheduling tools, and communicate clearly with frontline supervisors who did not care what the org chart said. The old move would have been obvious: write the requisition, post the job, buy distribution, and start sorting.

That is no longer the first move.

The first questions now are harder and much more valuable. Who already has 60% of these skills inside the company? Which adjacent employees could close the gap in 90 days? What internal projects, learning paths, or temporary assignments would make that move realistic? Only after that does the external market become the default answer.

This is the shift hidden inside the phrase “skills-based hiring.”

The first version of the idea was narrow. Stop overusing degrees. Search by skills instead of pedigree. Use assessments, structured interviews, and better data so candidates are judged by what they can do. That mattered. It still matters. LinkedIn’s June 2025 Skills Signal report says workers matched by skills rather than titles qualify for more than three times as many roles, and companies using skills-based talent searches are 12% more likely to make a high-quality hire. Adding 10 skills to a LinkedIn profile is associated with a one-month shorter employment gap.

Those are meaningful gains. They are not the whole story.

The deeper transition now underway is that skills are leaving the edges of recruiting and moving into the center of workforce operating logic. Skills are becoming the shared language through which companies decide whom to hire, whom to redeploy, whom to train, how to price roles, and where to place AI.

That is phase two.

It is less about screening candidates differently and more about running the entire talent system differently.

The name has not changed, which is part of the confusion. “Skills-based hiring” still sounds like a recruiting technique. In practice, it is becoming a data architecture question, a workforce planning question, and a budget allocation question. The companies that understand this are no longer asking whether they should remove degree requirements from job descriptions. They are asking whether they can build one skills layer that connects recruiting, internal mobility, learning, succession, scheduling, and service delivery.

That is a much bigger ambition. It is also where the real competitive value now sits.

Phase One Changed the Surface Faster Than the System

The first wave of skills-based hiring was necessary because the old proxies had become too blunt.

Degrees, employer brands, and linear job histories were once convenient shortcuts. They were cheap, scalable, and usually defensible to hiring managers under time pressure. Over time they also became increasingly detached from the way work was actually changing. Technical stacks moved faster than curricula. New roles appeared before job families could catch up. Workers learned through boot camps, project work, certifications, open-source contributions, community practice, and on-the-job adaptation rather than a single formal credential.

That reality pushed employers toward a more skills-first rhetoric.

Workers responded faster than companies did. LinkedIn’s 2025 Skills Signal report says the average number of skills added to member profiles rose 75% between 2019 and 2024, from five skills to nine. The probability of members adding digital skills in 2024 was 1.5 times the 2019 level. For green skills, it was twice as high. On the supply side, the labor market is already telling companies that skills signaling matters.

The employer side has been slower and messier.

Harvard Business School’s Project on Managing the Future of Work and the Burning Glass Institute published one of the clearest reality checks in 2024. Their report, Skills-Based Hiring: The Long Road from Pronouncements to Practice, found that the share of degree-required job postings had fallen only 3.6% over the relevant period. The researchers estimated that the net effect on actual incremental hiring of workers without degrees was just 0.14%, or roughly 97,000 workers annually out of 77 million yearly hires. Put differently, the much-advertised degree reset affected not even one in 700 hires. Nearly all of the measurable change came from just 37% of the firms in their sample.

That gap between language and behavior explains why phase one always felt simultaneously real and incomplete.

Job ads changed. Search filters changed. Assessment vendors grew. Employers learned to say “skills-first.” Yet when a hiring manager had to choose under uncertainty, the old heuristics often returned. The degree might no longer be mandatory in the posting, but it still functioned as a comfort signal in the interview room. The candidate might now be described in skills language, but the decision was still made inside a workflow built around jobs, titles, and manager intuition.

The difference between phase one and phase two is easiest to see in the table below.

Phase	Core promise	Unit of change	What improved	What stayed broken
Phase one	Hire for capabilities, not pedigree	Job post, search filter, assessment, interview process	More candidate access, better matching, broader pipelines	Internal talent visibility, learning links, workforce planning, manager incentives
Phase two	Use skills as the operating index for talent decisions	Whole talent system across hiring, mobility, learning, and planning	Better allocation, faster redeployment, stronger quality of hire logic	Governance, data quality, change management, ownership fights

Phase one made the labor market more legible. It did not make the enterprise more legible to itself.

That is the next problem companies are trying to solve.

AI Turned the Skills Debate Into an Operating Problem

For years, skills-based hiring could be treated as a progressive talent initiative, useful but optional. AI ended that luxury.

The reason is not simply that AI created new skills demands. It did. The reason is that AI also compressed administrative work inside recruiting while increasing the rate at which roles, tasks, and required capabilities change. Once both things happen at once, companies can no longer treat skills as a nice-to-have label on the edge of recruiting. Skills become the missing variable in day-to-day operating decisions.

The macro data is blunt.

The World Economic Forum’s Future of Jobs Report 2025 says job disruption will equal 22% of jobs by 2030, with 170 million new roles created, 92 million displaced, and a net increase of 78 million. Nearly 40% of skills required on the job are expected to change. Sixty-three percent of employers say skills gaps are the biggest barrier to business transformation.

LinkedIn’s Economic Graph work points in the same direction from a different angle. Its data shows that the skills required for jobs had already changed about 25% since 2015, and the number is expected to reach roughly two-thirds by 2030 as generative AI accelerates the pace of work redesign. In other words, the job may keep the same title while the skill content underneath it mutates steadily.

Recruiting teams are already feeling that pressure in workflow terms.

LinkedIn’s Future of Recruiting 2025 says 37% of organizations are now actively integrating or experimenting with generative AI in recruiting, up from 27% a year earlier. Among those organizations, the average time saved is about 20% of the work week. That is real operational relief. It means more drafting, summarizing, screening, and search work can be absorbed by software.

But saved time creates a harder question: saved for what?

If recruiters spend less time processing and more time making decisions, then the quality of the underlying decision system matters more, not less. A faster recruiting team that still cannot see adjacent internal talent, cannot connect skills to learning, and cannot tell whether the company already has the needed capabilities is simply making faster partial decisions.

Workday’s March 5, 2025 Global State of Skills release captures the pressure from management’s side. More than half of business leaders, 51%, say they are worried about future talent shortages. Only 32% are confident their organization has the skills needed for long-term success. Only 54% say they have a clear view of the skills inside their workforce today. Yet 81% agree that a skills-based approach drives productivity, innovation, and organizational agility, while 55% of organizations say the transition is already under way and another 23% plan to start this year.

That combination matters.

Leaders are worried. They lack visibility. They still believe the skills-based model is strategically necessary.

AI makes the gap more painful because it changes both sides of the equation at once:

it raises the premium on new and adjacent skills;
it automates more of the low-value work around talent decisions;
it makes it more expensive to rely on stale job architectures;
and it pushes companies to re-evaluate whether a skill shortage is truly an external hiring problem or an internal allocation problem.

This is why phase two feels different from the old degree-reset conversation.

The earlier debate was moral and methodological: should employers be fairer and smarter about how they evaluate candidates?

The new debate is operational: can a company still run on job-centric talent systems when work is changing faster than those systems can describe it?

Once that becomes the question, “skills-based hiring” is no longer just about recruiting. It becomes an enterprise control problem.

The Real Product Is a Shared Skills Index

When vendors talk about skills in 2025 and 2026, the most revealing phrase is not “skills-based hiring.” It is “single source of truth.”

Oracle states the thesis almost too clearly on its Dynamic Skills product page. It sells “a common skills language,” “skills-informed hiring,” “skills-driven learning,” “skills-informed career mobility,” “skills-driven scheduling,” and “skills-driven planning.” It also promises open skills aggregation, transactional data enrichment from recruiting and learning systems, and multilingual support across 29 languages.

That list tells you where the market is going.

The winning product is not just a better way to tag candidates. It is a shared translation layer between:

people and jobs,
jobs and projects,
projects and learning,
learning and mobility,
mobility and compensation,
compensation and workforce planning.

This is what a real skills index does. It takes capability data that used to live in separate systems and forces them into the same grammar.

SAP is making the same move from the suite side. In late 2024, Dan Beck described the objective behind SAP’s talent intelligence hub as infusing skills “throughout the entire talent journey,” directly from recruiting. By May 2025, SAP’s own messaging had expanded beyond talent acquisition into people intelligence, business data, workforce insights, and HR service delivery. On March 4, 2025, Caroline Hanke, who leads SAP’s skills-led organization work, put the internal logic plainly: companies should avoid looking externally for skills that already exist inside the firm but sit unused or in the wrong place. Transparency, she argued, is the point.

That sounds simple. It is not.

To make that transparency real, SAP is trying to connect skills data to hiring, learning, career development, and talent management. In a May 2025 talent acquisition piece, SAP highlighted Capgemini’s effort to build a skills engine spanning recruitment, onboarding, talent management, compensation and rewards, learning, and succession management. The point of the example was not that Capgemini had improved one recruiting workflow. It was that a single skills logic was being stretched across the employee lifecycle.

Workday’s framing is different, but the direction is the same. In February 2025 it launched Agent System of Record and explicitly listed Recruiting and Talent Mobility among its role-based agents. In that model, agents are managed not just as task bots but as role-based entities with configurable skills, policy controls, and governance. A month later, Workday’s skills research argued that traditional talent models based on job titles, degrees, and employer brands no longer offer enough visibility or agility. In June 2025, its recruiting marketing language went further, promising to reduce screening time, draw more requisitions from existing talent pools, and tie recruiting to internal mobility on one platform.

None of these vendors are using identical language. That is the important part. Different strategies are converging on the same architecture.

The common architecture looks like this:

create a normalized skills layer;
infer skills from profiles, work history, learning, and behavior;
connect that layer to open roles, internal gigs, mentoring, and development;
let AI recommend actions on top of that graph;
feed outcomes back into the system so the recommendations improve.

Once you see that pattern, the category becomes easier to read. The real product is not a requisition workflow with extra AI garnish. The real product is an index of who can do what, how well, how soon, and with what adjacent path.

That index is where the value compounds.

Without it, every talent decision becomes a fresh approximation. Recruiting teams write skills into job descriptions, learning teams map different skills to courses, mobility teams build separate opportunity catalogs, and workforce planners reason in another vocabulary again. The company spends millions on talent software and still cannot answer a basic question quickly: do we already have enough of this capability somewhere inside the organization?

Phase two exists because more employers now realize that the answer has to come from infrastructure, not intuition.

The Budget Is Moving Above Recruiting

The first buyer for skills-based hiring was usually a recruiting leader. The next buyer is more likely to be a CHRO, COO, or business leader who cares about workforce allocation across the whole company.

That is a structural change.

Recruiters still need skills data to search better and interview better. But once skills become useful for internal mobility, workforce planning, employee service, scheduling, and AI deployment, the budget conversation moves up and out. The spend starts competing less with other recruiting tools and more with broader HR and enterprise workflow investments.

LinkedIn has been signaling this shift for several years. Its data has shown that employees at companies with strong internal mobility stay materially longer; one widely cited 2023 analysis put the figure at 60% longer than at companies with low internal mobility. The same article noted that the skills required for jobs had already changed 25% since 2015 and were expected to keep moving sharply upward. Another LinkedIn internal recruiting guide says 69% of talent professionals believe internal recruiting accelerates new-hire productivity.

Those are not recruiting vanity metrics. They are balance-sheet metrics hiding in talent language.

Longer tenure protects retention economics. Faster productivity protects operating output. Internal mobility reduces the cost of external search, lowers onboarding drag, and preserves firm-specific context that outside hires lack.

This is where the phrase “skills-based hiring” starts to become slightly misleading.

What companies are increasingly buying is not just a hiring method. They are buying a way to allocate talent capital with less waste.

That shift is visible in vendor case studies because the examples are no longer framed only around recruiter efficiency. SAP’s May 2025 talent acquisition piece highlights Darussalam Assets reducing recruitment time from months to weeks and making hiring four times more efficient through AI-enabled workflows. Useful, yes. But SAP’s more strategic examples talk about skills engines, unified foundations, and planning across the employee lifecycle. The argument is not merely “your recruiters will save time.” It is “your enterprise will make better decisions about where work and talent should go.”

Oracle’s product language is even more explicit. Dynamic Skills is tied not only to recruiting and learning, but also to planning and scheduling. That means the skills conversation is already moving into operational domains where open roles, shift design, workforce capacity, and service delivery intersect. Once a platform can use the same skills layer to recommend who should apply for a role, who should take a course, and who should cover a shift, the buyer is no longer a single function.

Workday’s approach points the same way. Its skills research says AI helps organizations streamline repetitive tasks, enhance decision-making with data, personalize learning, and predict future skills needs. Those are enterprise management claims. They appeal to leaders who need to decide whether to hire, train, reassign, or redesign work.

That explains why the commercial center of gravity is shifting.

Phase one won support by promising that companies could widen the funnel and hire more fairly. Phase two wins budgets by promising that companies can reduce allocation error.

Allocation error takes several forms:

hiring outside for a capability the company already has internally;
training the wrong people for the wrong future role;
keeping workers trapped in narrow job families even though adjacent skills are transferable;
leaving managers blind to internal talent because systems only surface formal titles;
or using AI to accelerate decisions without the capability data to direct them well.

All of those errors cost money. All of them are more visible when finance is stricter, labor markets are less forgiving, and AI investment is forcing every function to defend spend with more precision.

That is why phase two is not a side topic in recruiting anymore. It sits much closer to workforce strategy.

The Bottleneck Is Governance, Not Good Intentions

The easiest mistake in this market is to think the hard part is taxonomy.

It is not.

A clean skills library matters. A common language matters. Parsing skills from resumes, job descriptions, learning histories, and work records matters. But most skills projects do not fail because the skill names are wrong. They fail because the surrounding organization never changes enough to trust and use the new data.

Workday’s March 2025 study is useful here because it asked leaders not only why they care about skills, but what stops them. The top barriers were not philosophical objections. They were execution problems: the time required to reskill employees, cited by 43% of leaders; resistance to change, 38%; lack of infrastructure, 28%; and inadequate skills measurement tools, also 28%.

That is governance language.

The HBS and Burning Glass findings point to the same underlying problem from the hiring side. Even after companies removed degree requirements from job ads, managers still used degrees as a practical risk heuristic. Why? Because under pressure they are choosing among imperfect alternatives, and the organization has not rebuilt the decision environment around them. If the company has not changed how managers evaluate evidence, which outcomes they are measured on, or how internal candidates are surfaced, then the system will drift back to familiar proxies.

Internal mobility illustrates the governance issue even more clearly.

Teuila Hanson, LinkedIn’s chief people officer, has argued that internal mobility works only when HR, talent acquisition, learning, compensation, and business leaders are at the table together. Rebecca Romano of NBCUniversal made a similar point at Talent Connect: programs and pathways do not matter much without a culture of access that gives employees real permission to move. Caroline Hanke’s SAP remarks land on the same operational truth from another side: external hiring should not be the reflex when internal skills exist but are hidden.

Each example points to the same constraint.

The technical system can recommend. The organization still has to allow movement.

That means phase two requires companies to solve several uncomfortable questions at once:

Who owns the skills record?

If recruiting owns it, the system becomes external-talent heavy. If learning owns it, the hiring connection weakens. If IT or a data team owns it, business adoption often lags. Someone has to define how skills are inferred, validated, updated, and challenged.

How frequently is the record refreshed?

A static skills profile becomes useless quickly. The good versions of these systems pull from learning completions, project work, job changes, performance signals, assessments, and manager or employee updates. The bad versions become graveyards of self-reported keywords.

What behavior gets rewarded?

Managers who hoard talent can quietly kill an internal mobility strategy. Recruiters measured only on external fills can do the same. Compensation teams that still price purely by job code rather than capability can turn a skills-based system into decoration.

Where does human judgment sit?

AI can recommend adjacent skills, infer likely fit, and suggest movement paths. It should not silently decide them. Skills systems work best when they improve decision quality while leaving room for manager context, employee ambition, and business nuance.

That is why governance matters more than enthusiasm.

Almost every large company now says some version of the same thing: jobs are changing, skills are becoming central, and AI is forcing a rethink. True. The harder question is whether the company is prepared to treat skills as living operational data rather than a one-time transformation project.

If the answer is no, phase two stalls. The organization keeps the language and misses the value.

The Vendors Are Fighting to Own the Index

Once skills become the shared layer across hiring, development, mobility, and planning, the competitive map changes.

The old question was which vendor had the best ATS, the best assessments, or the best internal marketplace.

The new question is simpler and bigger: who owns the index that everyone else has to plug into?

The main contenders are approaching that question from different starting points.

Player	Starting advantage	What they want to own	Strategic risk
LinkedIn	External talent graph and skills signaling at massive scale	Demand-side talent discovery, skills search, quality-of-hire signals, employer reputation	Strong outside view, weaker system-of-record position inside the enterprise
Workday	Core HR and financial system position in large enterprises	Managed skills record tied to recruiting, mobility, policy, AI agents, and budgeting	Needs to prove depth and usability, not just platform breadth
SAP	Integrated HCM suite plus business data context	Unified skills foundation across recruiting, learning, talent, service, and planning	Complexity and deployment burden can slow realized value
Oracle	Common skills language linked to planning, scheduling, recruiting, and mobility	Skills layer that connects talent data to broader workforce operations	Must convert infrastructure promise into day-to-day adoption

Each one is trying to answer a different version of the same enterprise question.

LinkedIn’s position is strongest when the problem starts outside the company: finding hidden talent, understanding external supply, expanding the candidate pool, and using skills signals instead of titles. Its own 2025 data says focusing on skills can expand AI talent pipelines by 8.2 times globally. That is a powerful argument at the sourcing edge.

Workday, SAP, and Oracle are stronger when the problem starts inside the company: how to connect the recruiting workflow to the employee record, the learning catalog, the mobility marketplace, and the planning model. Their pitch is not merely that they know what skills matter. It is that they can connect those skills to actual enterprise action.

That distinction will shape the next round of category winners.

Point solutions can still win if they do one thing much better than the suites. Assessment platforms can generate better signal. Talent intelligence vendors can model skills more deeply. Internal marketplace specialists can create stronger mobility behavior. But their long-term defensibility depends on whether they become indispensable inputs to the index or just nice features around it.

The market has already started to separate the two.

If a product improves one part of the recruiting process but cannot export trusted skills data into the rest of the talent stack, its value is narrower than it first appears. If a product becomes the place where recruiting, learning, mobility, and planning all need to read and write capability data, it sits much closer to the enterprise wallet.

That is why the category feels less like HR tech in the old sense and more like infrastructure.

The battle is no longer only about who can find the best candidate. It is about who can let a company answer, in one decision flow:

what capability is actually needed,
where that capability already exists,
which adjacent workers can reach it,
what learning closes the gap,
what must still be bought externally,
and how AI should support the whole process without turning it opaque.

The vendor that owns that flow owns far more than recruiting.

The Job Ad Becomes the Last Step, Not the First

The most important thing to understand about phase two is that it does not eliminate hiring. It changes where hiring begins.

In the older model, a role opened and the system looked outward first. The job ad was the trigger. Skills were attached later as descriptive metadata.

In the newer model, the role opens and the system looks across capabilities first. Skills become the starting index. The external market is still there, sometimes decisively so, but it becomes one channel among several rather than the automatic first step.

That shift sounds procedural. It is strategic.

It changes how companies think about scarcity. A skill shortage may still be real, but phase two forces a more disciplined question: is this a shortage in the external market, or a shortage in our visibility, mobility, and training systems?

The answer will often be uncomfortable. Many companies have more capability than they can see and less coordination than they imagined. AI has made that harder to ignore because it accelerates routine recruiting work while making role content more fluid. Once routine work gets cheaper, the real cost sits in bad allocation.

That is why the second generation of skills-based hiring matters more than the first.

The first generation tried to improve who got considered.

The second is trying to improve how the enterprise decides what talent problem it actually has.

That is a bigger promise, a harder implementation problem, and a much more valuable market.

The next time a talent leader opens a role, the most revealing click will not be “post job.”

It will be “show me the skills.”

This article provides a deep analysis of how skills-based hiring is evolving from a recruiting method into a cross-system talent infrastructure. Published April 13, 2026.