The Broken First Rung: Who Creates Early-Career Experience?

For decades, enterprises operated with a relatively simple assumption. Universities developed foundational capability, employers hired graduates, and the first few years of work converted education into professional competence. Nobody wrote that contract down, but every campus program, graduate scheme and analyst rotation was built on it.

That contract is changing. Our analysis of entry-level and early-career job demand between August 2025 and August 2026 shows that employers increasingly want people at the beginning of their careers to arrive partially trained. One or two years of experience is frequently treated as the starting point. Technical fluency is moving into non-technical industries, AI tools are appearing in junior job descriptions, and regulatory and domain knowledge is being pushed toward the front of the career.

Yet many enterprise talent systems still treat entry-level as a recruiting category rather than a workforce architecture. That creates a structural problem, and it is worth stating plainly: if every employer wants someone who has already completed the first one or two years of development, who creates that experience? The message for the CHRO is straightforward. Do not compete only for early-career talent. Build it.

The first rung has moved, and the gap below it has no owner

The most consequential finding is not a shortage of postings. It is the shape of the experience curve. Postings explicitly requiring zero years of experience number 707,751. Postings requiring one year number 11.6 million, roughly sixteen times the true zero-experience band; two years accounts for 11.2 million and three years for 9.7 million, which makes three years the statistical midpoint of the market. The explicit gateway into professional work is small, and the new center of entry demand sits above it.

Inside a single requisition, the logic is sound. Experienced candidates ramp faster, hiring managers perceive lower execution risk, teams have less capacity for apprenticeship than they used to, automation is eroding the repetitive work historically used to develop new employees, and specialized workflows reward candidates who understand the domain before Day One.

The problem appears only when many enterprises behave this way at once. Every company wants to recruit people after someone else has trained them, and the result is a structural gap between zero and one year that nobody owns. Requiring a year or two feels like risk reduction inside the requisition; across the labor market it moves development responsibility elsewhere. That should change the economics of early-career programs. Internships, apprenticeships and rotations are mechanisms for manufacturing a scarce input, not employer-brand initiatives.

Two talent markets are hiding behind one label

The second finding is that entry-level now describes two fundamentally different markets. One is operational and volume-led: retail associates, warehouse roles, customer support. The other is professional and skills-intensive: junior accountants, software engineers, data analysts, financial analysts. They differ in work design, skill requirements, sourcing channels and career economics, and treating them as a single graduate population obscures all of it.

The operational market runs on workflow execution and operational systems, and its development challenge is scale and consistency. The professional market runs on technical fluency, analysis and domain knowledge, and its challenge is speed to productivity. The risks differ too: treating an operational role as a bridge when it may not build transferable experience, or requiring prior experience in a professional role while underinvesting in creating it.

Work arrangements reinforce the divide: early-career postings carry a higher part-time share than the broader market, 14% against 11%, and less contract work, 9% against 14%. Not all experience compounds equally. A candidate can accumulate time in an operational role without acquiring the technical or workflow experience a professional early-career role requires. Stop assuming that any first job creates a natural pathway to the next. The bridge has to be designed.

The portable core is digital; the differentiating layer is not

Across the nine industries we examined in skill detail, only two technical skills appear consistently in the top 100 everywhere: Python and SQL. The rest of the contract diverges sharply. HIPAA and patient care in Healthcare. KYC and credit analysis in Banking. Merchandising and loss prevention in Retail. BGP and fiber optics in Telecom. FMEA and APQP in Automotive. Pharmacovigilance and regulatory submissions in Pharma.

Underneath the divergence there is a common direction. Every sector is building a digital layer on top of its traditional domain workforce: analytics on clinical work, quantitative skills on risk and relationship work, data capability on merchandising, computational capability on science, software and AI on network engineering. Automotive is adding software to mechanical engineering, creating a dual engineering labor market in which a manufacturer now competes with software companies for part of its early-career workforce.

None of this means Python and SQL should be mandatory for a cashier, a nurse aide or a warehouse associate. It means the optimal profile for professional early-career roles is T-shaped: portable computational literacy plus deep domain context — clinical workflow and data in Healthcare, risk and SQL in Banking, science and computational analysis in Pharma, mechanical engineering and embedded software in Automotive. That is where university partnerships should move next, away from "send us graduates in this major" and toward the combinations of domain and digital capability the enterprise will need in three years.

Readiness has moved to before Day One

Early-career postings emphasize communication, organization, collaboration and time management, with attention to detail, professional ethics, self-direction, ownership and dependability appearing prominently. AI-related skills appear in meaningful volumes in junior postings. The implication is not that every graduate must be an AI engineer. It is that enterprises increasingly expect new entrants to arrive knowing how to work productively in an AI-enabled environment.

The right response is not a generic two-hour AI course for graduates. It is a common foundation — what models can and cannot do, enterprise-approved tools, prompt and workflow design, validation, data privacy, responsible use — followed by role application: coding acceleration for engineers, research for finance, clinical workflow support in healthcare. The value is not AI vocabulary. It is AI-enabled productivity inside the work.

The soft-skill contract has changed in the same direction: reliability before leadership. As routine work becomes easier to automate, the human contribution moves toward judgment, verification, exception handling and accountability. Early-career talent needs less organizational authority than it once did, but potentially more judgment, earlier. Assessment should follow. Replace "shows leadership potential" with observable evidence: does the candidate follow through, identify errors, communicate uncertainty, verify AI-generated outputs and handle operational exceptions? That is a more useful definition of early-career potential, and a more assessable one.

Design the 0-to-3-year workforce as a system

Four decisions are worth making inside a quarter. Audit every entry-level requisition by asking what would actually fail if the candidate had zero years of experience; where the answer is a skill that can be assessed directly or taught in a structured ramp, reconsider the requirement. Split the 0-to-3-year workforce into explicit populations: true entry, transitional early career with internship or apprenticeship exposure, professional early career, and credentialed entry where qualification precedes employment. Identify the roles where you should manufacture experience rather than buy it — where external talent is expensive, the capability is teachable, demand recurs and retention economics justify the investment. And change the success metric: time to independent productivity, skill attainment at six and twelve months, retention at twelve and twenty-four months, and cost to build against cost to buy comparable two-to-three-year talent. Not the number of graduates hired.

Timing is a planning variable, not a scheduling detail. The twenty-four-month series shows meaningful seasonality: March 2026 reached roughly 565,000 early-career postings while December 2024 was roughly 292,000, a peak-to-trough difference of about 1.9 times. February through April is consistently strong, with a secondary window around September and October. The implication is not to recruit in March. It is to align university engagement, assessment, internship conversion, offers and start dates with actual market competition rather than an inherited campus calendar.

Job-posting data measures what employers ask for, not what employees possess or how they perform, and the source analyses are global rather than US-only, so read the numbers directionally. The structural point holds whatever filter you apply: the first job is increasingly asking for experience, and someone has to create it. The enterprises that own part of that experience curve will pay less later for talent developed by everyone else.