Workday's Mobility Agent Meets a Hiring Pause

AI-generated editorial illustration.

On October 5, Workday put two uncomfortable numbers in its workforce report. Mentions of hands-on AI building skills in job requisitions rose 51% from September 2025 to July 2026. Internal moves fell at 57% of the employers in a separate, matched year-over-year customer sample. The first number describes advertised demand. The second describes career movement inside companies. Neither says AI caused the decline. A buyer weighing a career-matching contract now has to ask what happens when the software finds a worker who can do a job that the company will not approve.

Workday has a direct commercial interest in the answer. Its Talent Mobility Agent promises to identify relevant openings from a worker’s skills, experience and interests, then bring opportunities into email or Microsoft Teams. Workday also sells Career Hub, Talent Marketplace and Skills Cloud. Those products may make the route to an opportunity easier to find. They cannot create a funded requisition, compel a manager to release a strong employee or make an opaque selection fair.

The new workforce release does not test the mobility agent against the customer base. It combines several kinds of evidence: de-identified HR records from active customers with at least 250 employees and a matched prior year; roughly 550 Workday Recruiting employers’ requisition data; a global survey of 6,001 workers and leaders; and a separate 5,944-person AI survey. A buyer who collapses them into one before-and-after chart will assign precision the report does not have.

The release describes a mismatch a dashboard cannot repair on its own: skills in open requisitions change faster than workers move into those jobs. Workday can sell the route across that gap. The employer controls the doors at either end.

October 5 split skills demand from career movement

Workday’s headline interpretation is that AI is rewriting jobs more than removing them. In its survey, 40% of business leaders expected AI to help them get more from existing employees, while 28% expected it to reduce headcount. These are expectations, not observed staffing changes. Nearly four in ten employees said their company had reorganized or restructured in the prior year. That is an experience report, not an estimate of jobs eliminated because of AI.

The distinction matters to a worker whose job description changes while their pay band and title remain fixed. A manager may add automation design, tool validation or process ownership without opening a higher-level post. A requisition can ask for new skills even if the internal ladder does not gain a rung. Workday said global promotion rates in its customer data were essentially flat. Some people still advanced. The published figures simply do not show promotion rates rising with the new skill vocabulary.

The 57% figure has a narrow denominator. It is the share of employers in Workday’s matched customer records where internal moves fell year over year. It is not a 57% drop in the number of moves. It is not 57% of workers. The public release does not provide the distribution of decline by firm size, industry or AI adoption. Nor does it show whether the employers buying Workday’s mobility tools were more or less likely to see a fall. A company could increase its own internal moves even as most peers in this sample saw fewer. The reverse is also possible.

A lower formal-move count has another possible reading. During a hiring pause, a company may put employees on cross-functional projects instead of creating permanent posts. A team may redesign an incumbent’s job without changing the payroll title. Both choices can preserve employment and build useful experience. Workday’s headline measure cannot show whether that happened. To make the favorable case, an employer would need records of project access, paid time, completed work, recognition in later hiring or promotion, and a route out if a temporary assignment never becomes a job. Otherwise fewer formal moves could mean more work with less advancement.

Workday’s survey adds the human obstacle. Employees trying to move internally most often reported paused hiring for the role, managers who did not support the move, and selection that did not feel fair. About half said they had not even tried an internal move in the previous year. Among those non-applicants, 27% said they saw no attractive opening inside their company. These are different failure points. A poor match can suppress an application; a paused requisition can make a good match useless; a manager can prevent a move after a candidate has qualified.

The HR record may show all three as one employee who did not transfer. Yet they belong to different decision makers. The platform may help the worker discover an opening. Finance and operating leaders approve the seat. The sending manager must release the worker; the receiving manager decides whether to hire. A zero at the end of that sequence cannot tell a buyer which decision failed.

A 51% gain does not mean a promotion

The release says mentions of basic AI skills, such as simple prompting, climbed through late 2025, peaked in January 2026 and then fell 25% over the following months. Mentions of applied skills, including building AI tools, automating workflows and AI engineering, rose 51% between September 2025 and July 2026. The population is roughly 550 enterprise employers that use Workday Recruiting. Those figures describe the frequency of skills in their requisitions. They do not establish a 51% increase in vacancies, a wage premium for builders or a 25% fall in the value of prompt literacy.

Workday’s public summary does not provide the underlying job count or an occupational breakdown for these changes. An employer can also rewrite a job ad without changing its staffing plan. The rise across hundreds of enterprises is a demand signal, not a payroll result. A company should compare its own requisitions with project assignments and pay decisions before using the global percentage to budget for a department.

Another pair of figures suggests that workers already recognize the change. In Workday’s September survey, 79% of workers said they know which skills they need to succeed, while 66% said their employer helps them develop those skills. The 13-point difference is a support gap in survey answers; it is not a measured shortage of training places. Sixty-five percent said they were confident they could learn new skills if asked. That confidence makes the absence of a credible route into new work harder for an employer to explain.

Workday’s separate August AI@Work Pulse found 62% of heavy AI users thought AI would make their current skills less valuable, while 76% expected it to open new career opportunities. Those beliefs can coexist in the same person. A worker may see routine work losing value and still expect a chance to design or supervise an automated process. But an opportunity is not an internal offer. The employer must define the work, decide whether it belongs in a new role and give someone time to learn it.

The report also found fewer mentions of management and leadership skills, down 7%, and training skills, down 13%, in requisitions over the same September-to-July period. That does not mean companies no longer need managers or trainers. It does raise a practical question about the labor needed to make new AI skills useful. Someone has to coach a worker, review a project, support a transfer and cover the vacancy the transfer leaves behind. Those hours rarely appear in a product demo.

The employer’s expense extends beyond a course or license. A real transfer can require protected practice time, an assignment where the skill is used, an assessment, the receiving team’s onboarding and a replacement plan for the sending team. Workday’s release prices none of these. Course completions make an easy slide. They leave out the labor that turns learning into a changed job.

A match still needs an approved role

First, someone must put a real opportunity into the system. It may be a full-time opening, a short project or a flex team assignment. Its existence depends on decisions made outside a matching model. Workday’s Career Hub administrator guide says employers can enable suggestions for internal jobs, flex teams and projects. The Opportunity Marketplace uses skills-based matching and can recommend learning to close gaps. The setup guidance also says employers have to configure the relevant products and job architecture. Poorly described job profiles or missing skill qualifications reduce the quality of suggestions.

The hardest missing field can be simpler: approved capacity. A company may know it needs an AI workflow builder and still be holding headcount flat. It may have a project but no position to transfer the employee into afterward. A department may advertise a role while its manager intends to fill it externally. A matching system can infer adjacent skills; it cannot decide which budget bears the salary. These are operating possibilities, not incidents Workday’s report documented at a named customer.

The administrator setup instructions show how much of the experience is a customer decision. An employer chooses whether the marketplace suggests jobs, projects or flex teams, supplies the corresponding requisition or project data, and configures security and worker profiles. It can also restrict suggested jobs to a worker’s country. If internal jobs are not enabled, a worker may see a promising project but no permanent opening in that product surface. If a job profile lacks relevant skills, a worker with experience in another department may not appear as a good match. The vendor supplies the matching system; the customer decides what the system is allowed to see and show.

In the public survey, employees named hiring pauses and manager resistance among the barriers they experienced. Opening more courses while leaving those rules untouched can expand the number of employees who are qualified and stuck. That is especially acute where training is being sold as a retention measure. A worker may spend months building a capability, discover a matched role and then learn that the requisition is frozen. The calendar cost is real even if the HR system records the outcome simply as no move.

Outside the company, the queue is longer. Workday said the median applications per job filled rose from 58 to 69 over one year, while the time to fill stayed around 60 days. In its AI survey, 84% of job seekers said they used AI in their search. These figures do not prove that AI created all the extra applications or slowed hiring. They do show why telling a disappointed internal candidate to look outside is a costly fallback: a bigger application pile does not automatically deliver a faster offer.

The U.S. Bureau of Labor Statistics’ August 2026 JOLTS release reported about 7.1 million openings and 5.2 million hires, both little changed from July. JOLTS covers U.S. establishments and counts openings and flows across the wider labor market. It cannot be placed on the same axis as Workday’s global customer mobility figures. It does provide a reminder that neither abundant postings nor a skills shortage slogan automatically translates into completed transitions.

AI-generated editorial illustration. A visible skill path can stop before an approved job while another path leads to project work.

Workday sells the route from profile to opening

An employee does not have to visit a separate jobs site to receive every suggestion. Workday’s Talent Mobility Agent is built to bring internal career decisions into existing work. According to the product page, it can align workers with openings using skills, experience, interests and a company’s mobility rules. It can surface opportunities by email or Microsoft Teams, aid recruiters in rediscovering internal candidates and prompt managers to support movement. The page also lists a dashboard for tracking internal mobility. A separate Workday guide describes Career Hub, Talent Marketplace and short-term Flex Teams as different capabilities. Availability and setup depend on what the employer enables.

The product addresses a real information problem. As an illustrative scenario, not a reported customer case, an engineer in one unit might not know that a finance operations team is building an automation project. A recruiter might overlook that employee because their current title does not match the new requisition. Skills, interests and project history can reveal an adjacency that a title-only search misses. A notification at the right time may get the worker to apply before the position is filled. None of these benefits requires claiming that an algorithm should make the hiring decision.

For a manager, however, that improved discovery can feel like an immediate vacancy on their own team. Releasing an employee may mean losing the person who knows a customer account or keeps an old workflow running. A reasonable mobility policy has to make that cost visible and assign a transition plan. It might set a release window, budget a short overlap between teams or let the employee work on a trial project before a permanent move. Workday’s survey reports unsupportive managers as an employee experience. It does not say why each manager refused, or whether a replacement was available. Treating every refusal as irrational would miss the budget constraint that made the software recommendation hard to act on.

Workday points to an unnamed Fortune 500 professional services customer’s results: 30% more internal applications through AI notifications, 1.4 times more higher-quality internal applicants than external applicants, 2.3 times greater likelihood of moving to a new role and a 5% improvement in employee retention. These are vendor-presented case figures, not the October report’s overall results. The public page does not expose a matched control group, baseline period or definition of higher quality for that case. The figures suggest that better discovery can matter where a customer has usable openings and processes. They do not establish what an average buyer will receive.

Workday’s Skills Cloud page reports a 26% higher internal hire rate and 5% lower voluntary turnover in a product-level analysis of customers where data were available from July 2023 through June 2025. The sample and period differ from the October 2026 workforce report. Treating the earlier product association as a rebuttal to the later decline would be just as misleading as treating the decline as proof the product failed. Adoption, employer selection, hiring policies and labor conditions are not held constant in the public summaries.

Workday’s matching features can recommend a job or a short-term project. The project may be valuable when a permanent role is frozen: a worker can demonstrate an applied skill while a team tests work before adding headcount. But a project without paid time, manager permission, evaluation or a route into a role can become a substitute for advancement. The same recommendation may widen opportunity for one employee and add invisible work for another. The employer’s rules, not the graphic quality of the match, decide which one it becomes.

LinkedIn offers a rival path through the same bottleneck

LinkedIn is selling its own career infrastructure. Its Learning Career Hub page says organizations using the product show 20% higher internal mobility rates and 22% longer tenure; it also cites 3.4 times faster AI skill growth among LinkedIn Learning users. Those are vendor claims on selected users, not a randomized comparison to employers in Workday’s report. They make the buyer’s question sharper: which action changed after the system showed a worker a path?

LinkedIn’s 2026 Future of Talent report describes a wider operating model in which talent teams link external hiring, internal mobility, learning and workforce planning. Its June survey covered 771 HR and talent acquisition leaders at employers with more than 1,000 employees in six countries. The report says more than a third of its defined frontier group expect recruiters to support retention, mobility and redeployment. That is a prediction from a selected subset of respondents, not an observed mobility rate. Yet the role shift is concrete: a recruiter who can propose an internal candidate before a requisition is finalized may change the choice between retraining and external search.

The same report describes Paycom’s move toward calling recruiters talent advisers and HSBC’s combination of talent acquisition and talent management in one function. These named examples are not proof that every company’s recruiters will take ownership of internal careers. They show where a budget and reporting line might move. If a recruiter is judged only by external fills, an internal referral can look like work that does not count. If the talent team can count a redeployment against a business vacancy, it has a reason to help a worker move before a hiring manager opens the external funnel.

LinkedIn gives another operational example: PepsiCo says standardized processes across its talent teams underpin a Talent Foresights capability for modeling future workforce needs. In the same report, HubSpot vice president of talent and people experience Becky McCullough frames the question as whether to grow, hire or develop in service of a business goal. These are company descriptions carried in LinkedIn’s report, not independent outcome audits. They matter because a mobility system has to compete with other ways of filling work. A finance leader could fund a new external hire, permit an internal transfer and backfill the old team, or ask the current team to stretch. The product can improve the information available for that choice. It cannot make the tradeoff disappear.

The OECD’s June 2026 skills-first report draws a useful boundary. It argues that skills development and skills recognition have to move together. A person can learn a skill, but employers still have to record, assess and value it in hiring and progression. The report says qualification requirements, professional experience and organizational norms still shape career decisions, and removing a degree requirement alone does not broaden access. That is a structural counterweight to the promise that better AI matching by itself will unlock a career ladder.

Workday, LinkedIn and the OECD examine different parts of this problem. The vendor products organize profiles, suggestions and applications. The OECD examines the institutions that recognize competence and connect learning to work. A high match rate can coexist with a low move rate. A buyer needs to ask where the two connect.

A mobility conversion ledger separates signals from moves

An employer cannot diagnose an internal career freeze from a single dashboard total. Start with a defined period, eligible employee population, approved opportunities and a comparison group or historical baseline. Separate jobs from temporary projects and learning recommendations. Then follow the worker through each decision. The table below is an editorial measurement design, not data Workday or LinkedIn reported for a customer.

Stage	Count and denominator	Owner and question
Eligible workers	Employees permitted to apply internally, by tenure, function and location	HR: who is excluded before matching begins?
Approved opportunities	Funded internal jobs open for applications, plus projects in a separate column	Finance and operating leaders: how many real seats exist?
Relevant suggestions	Workers shown an opportunity among those eligible, with an auditable match rule	Product team: who is visible, and whose skills are missing?
Internal applications	Applicants per visible, approved job, split by cohort and seniority	Talent team: does discovery turn into a candidate decision?
Manager release	Qualified applicants allowed time and permission to move	Sending manager: where do approvals stall?
Offers and moves	Offers accepted and payroll-recorded role changes per internal applicant	Hiring manager and HR: who actually changes role and pay?
Durable outcome	Six- and twelve-month retention, performance and pay after the move	HR and Finance: did the transition last and justify its cost?

This sequence prevents an easy reporting error. A vendor can raise applications while the company lowers approved requisitions. The application rate may improve even as the number of moves falls. Both statements can be true. The ledger should show counts and rates, plus the date and source system for each, so a procurement review can see whether the bottleneck is discovery, candidate qualification, manager release, budget or selection.

The time window matters. A worker might receive a suggestion in one quarter, complete a project in the next, and move into a permanent role months later. A 30-day conversion report would call that route a failure; an annual report could credit the software for a move that began elsewhere. Track the same cohort over a stated period, record when a job was actually approved, and disclose employees who leave the company before a move. Split promotions from lateral transfers. Otherwise an employer could report more internal mobility while workers see no change in responsibility or pay, or report less while valuable project assignments increase. The ledger is meant to make those interpretations testable.

Fairness belongs in the same report. If managers decide who is allowed to apply, a matching model may surface jobs only to workers already favored by a manager. If development projects happen outside paid hours, caregivers and shift workers face a different price of entry. A company should compare visibility, applications, release, offers and pay outcomes across relevant groups, within applicable privacy and employment rules. The October Workday survey gives a reason to ask: employees cited selection that did not feel fair. It does not tell us which employers had an unlawful process, and this article cannot infer that from a survey answer.

The ledger also changes what an AI skills program can claim. A course completion shows learning activity. A skills assessment shows a narrower capability. An approved role or project shows demand. A payroll move and new assignment show deployment. The costs of backfilling the old team, onboarding the new team and supporting a worker through the transition belong in the return calculation. The same caution applies to employees: acquiring a sought-after skill can improve options, but no public report promises a specific internal promotion.

A requisition review assigns the decision

Workday’s October report leaves a paradox in the buyer’s lap. In one sample, companies are asking for more applied AI skills. In another, most employers saw fewer internal moves. The data do not say which firm used a mobility agent, who held a hiring pause or how many workers moved because of a recommendation. It would be wrong to turn the parallel lines into a causal chart.

It would be equally wrong to ignore the operational gap. Workday and LinkedIn can make a hidden skill or opportunity visible. Their selected customer cases suggest that some employers convert better discovery into more internal movement. But a worker’s career does not happen in the recommendation pane. It happens when a position is funded, a manager releases time, another manager accepts the person and the payroll record changes. The gap between those events is where training budgets and software claims should be tested.

At the next requisition review, the company can ask for three numbers before approving another mobility feature: qualified employees who saw the opening, qualified employees whose managers supported an application, and employees who moved into the job. If the offer went outside, the hiring manager can explain the decision against those records. The skill recommendation has already done its work. The career move still needs someone to sign it.