# Learning and Development Work in the United States: Employer Demand and Practice in 2026

> A dated study of 110 U.S. employee-learning vacancies examines analysis, learning plans, facilitation, assessment, evaluation and improvement.

- Canonical page: https://mtfinstitute.com/insights/learning-development-us-employer-demand-practice-2026/
- Content type: Article
- Editorial category: Research & Reports
- Publisher: MTF Institute of Management, Technology and Finance
- Author: MTF Institute Research Team- Published: 2026-09-04
- Updated: 2026-09-05
- Language: English
- Topics: Learning and Development, Learning Design, Assessment, Vacancy Analysis, Employee Training, Needs Analysis, Facilitation, Evaluation

## Learning and Development Work in the United States: Employer Demand and Practice in 2026

The report and its eight-file research archive are available at [Zenodo DOI 10.5281/zenodo.22313254](https://doi.org/10.5281/zenodo.22313254). [Download the research report PDF](https://zenodo.org/records/22313254/files/learning-development-us-employer-demand-practice-2026.pdf?download=1).

**Author:** MTF Institute Research Team  
**Institution:** MTF Institute  
**Publication date:** 5 September 2026  
**Research institution report number:** MTF-CF-RR-2026-09-05-29  
**DOI:** 10.5281/zenodo.22313254

## Abstract

What do United States employers ask learning and development (L&D) practitioners to do in 2026? This report examines 110 publicly accessible U.S. vacancies observed on 4 September 2026. The sample was purposive and nonprobability based: it was assembled to obtain a varied, auditable view of employee-learning work, rather than to estimate the national share of vacancies containing a requirement. Sixty-eight records came from first-party employer or public-sector pages and 42 from readable full-description job-board pages. Every record was coded for six non-exclusive work areas: analysis of needs or performance gaps, learning plans, facilitation, learner assessment, programme evaluation, and evidence-led improvement.

Learning plans appeared explicitly in 103 of 110 records, programme evaluation in 93, improvement in 91, analysis in 89, facilitation in 78, and learner assessment in 47. These are sample description counts, not estimates of national prevalence. Non-mention does not establish that a capability is unnecessary. The findings show a connected professional workflow: practitioners translate a workplace problem into a learning response, plan experiences and resources, support practice, check learning, examine programme results, and revise. They also show why learner assessment and programme evaluation must remain separate decisions. A knowledge check can show whether a person learned something; it does not by itself show whether the programme improved work.

The vacancy evidence is read alongside U.S. occupational and public-sector guidance. The U.S. Bureau of Labor Statistics reports a 2025 employment base of 472,500 training and development specialists, projected growth of 11% from 2025 to 2035, and about 46,000 openings per year on average, including replacement needs. That national forecast is distinct from this one-day vacancy sample and does not forecast demand for any course. The report concludes with practical implications for employers, practitioners, and professional education.

## 1. Research question and scope

The study asks: what kinds of work were explicitly described in a purposive sample of current U.S. vacancies involving employee learning and development? It focuses on work done for employees or client workforces. It excludes school and student teaching, consumer-only education, recruitment-only activity, roles that merely receive training, machine-learning engineering, and clinical instruction requiring professional licensure.

This boundary matters because “learning” appears in many kinds of employment. An instructional designer may create employee onboarding, customer product education, university courses, or school materials. A trainer may support warehouse operations, software use, leadership practice, or regulated technical work. The study admits a vacancy only when the readable duties support workplace employee development and an identified U.S. location. It uses duties and learner population, rather than title alone, to decide fit.

The evidence describes employer requests at a moment in time. It does not measure how incumbents spend their time, whether a vacancy was later filled, the effectiveness of an employer's programme, or the employment effect of a qualification. Senior and specialist vacancies illuminate the workflow and its constraints, but do not imply that an introductory course qualifies a beginner for those posts. Technical, safety, medical-device, aviation, defence, and similar roles may require domain knowledge or credentials beyond general L&D practice.

## 2. Methods

### 2.1 Sampling design

The research geography was declared as the United States before collection. Researchers searched publicly accessible employer and applicant-tracking pages and readable full-description job boards for roles involving corporate learning and development, employee training, instructional design for employees, learning programmes, onboarding, facilitation, and related work. Collection took place on 4 September 2026.

The final corpus contains 110 unique records. Sixty-eight, or 61.8%, are in the first-party stratum: employer-controlled careers or applicant-tracking pages and one public-sector employer bulletin. Forty-two, or 38.2%, are board fallbacks with readable full descriptions. The source split is disclosed because a board can preserve a useful description while providing less direct evidence of employer-controlled status or wording. Board records were not represented as independently employer-confirmed vacancies.

The design is purposive, nonprobability sampling. It intentionally sought variation in sectors, role families, and seniority while favouring sources whose full duties and current-status signals could be inspected. It is affected by searchability, public access, ATS rendering, English-language availability, and the sectors and employers returned by the searches. It has no known selection probability, complete sampling frame, or exhaustive search denominator. Percentages therefore describe this corpus only.

### 2.2 Eligibility and current status

Records required readable duties, an explicit U.S. location or eligibility basis, a substantive employee-learning component, and credible observable current status. An application control alone was insufficient if the page also said that the post was closed or expired. Search snippets, redirected generic boards, unreadable shells, non-U.S. positions, elapsed closing dates, and ambiguous learner populations were excluded.

Where a source exposed an exact publication or closing date, it was retained. Where it did not, the researchers kept the retrieval date and the observable status evidence. A crawl date, “new” label, or relative board age was not silently converted into an employer posting date. This creates a real freshness limitation: a public page may still accept an application while the employer has privately paused or filled the role, and a board may retain an apparently active description after employer conditions change.

### 2.3 Deduplication

The researchers counted one requisition rather than each location, shift, or potential headcount. URLs were normalised by host and path with tracking parameters removed, while original observed URLs were retained. Stable requisition identifiers were preferred. Employer, requisition, title, and location were compared to identify reposts and near-duplicates. Multiple locations attached to one requisition were collapsed; distinct requisitions were not merged merely because their titles matched.

The combined source collection contained 114 candidates before final selection. Four were excluded during review: one reached its stated close boundary; two did not establish an employee learner population; and one returned a first-party “job not found” page on 5 September. The resulting 110-record dataset passed checks for unique record identifiers, normalised URLs, employer-requisition pairs, U.S. geography, six-category values, and parity between the structured source and tabular export. The public accepted-vacancies CSV has SHA-256 `11b4416d3cda94fff3e48bd90f783768232ff2d64f6442e05cc5469457bb2634`.

### 2.4 Coding

Each vacancy was coded in six non-exclusive categories. Analysis meant explicit investigation of learning or performance needs, task requirements, learner gaps, or stakeholder requirements. Learning plans included objectives, programmes, sessions, pathways, schedules, resources, or delivery plans. Facilitation required delivery of employee learning, workshops, guided practice, facilitator coaching, or live learning discussion; generic presentation skill did not count.

Learner assessment meant checking a learner's knowledge, skill, demonstration, qualification, or performance against defined criteria. It excluded assessment of job applicants. Programme evaluation meant examining learning outcomes, effectiveness, transfer, or relevant learning metrics. An employer's evaluation of the candidate did not count. Improvement meant revising learning content, delivery, or programmes using feedback, evidence, or changing needs. Generic organisational language about continuous improvement did not count.

A category received a positive code only when the readable description supported it. A zero means “not explicitly observed in this description,” not “absent from the job” and not “unnecessary in practice.” Categories overlap, so their counts do not sum to 110. Coding bases were short original paraphrases; retained extracts were minimal and rights-limited.

### 2.5 Review and rights

The evidence received a targeted independent audit rather than full duplicate coding of every record. Independent review examined a spread of retained and excluded records, disputed categories, source status, geography, duplicates, and every material numeric claim. Corrections identified by the audit were incorporated before the corpus was frozen for analysis. Because the whole corpus was not independently double-coded, no inter-rater reliability score was calculated or claimed.

The retained evidence consists of bibliographic facts, URLs, status notes, concise original coding bases, and minimal extracts. It does not preserve bulk vacancy text, proprietary models, questionnaires, paid tables, or protected frameworks. This report uses original synthesis and does not reproduce employer descriptions.

## 3. Findings

### 3.1 Six areas of L&D work

| Work area | Explicit mentions | Share of 110-record sample |
|---|---:|---:|
| Learning plans | 103 | 93.6% |
| Programme evaluation | 93 | 84.5% |
| Improvement | 91 | 82.7% |
| Analysis | 89 | 80.9% |
| Facilitation | 78 | 70.9% |
| Learner assessment | 47 | 42.7% |

Planning is the most widely observed area. In practice, the category covers more than writing a course. Employers describe the coordination of programmes and sessions, definition of objectives, design of learning resources, selection of delivery arrangements, management of pathways, and alignment with stakeholders. The very high count, 103 of 110, suggests that the sampled work is usually organised around deliberate learning provision rather than isolated presentation activity.

Programme evaluation and improvement also appear frequently: 93 and 91 records respectively. Their proximity matters. Many roles ask practitioners to monitor effectiveness, outcomes, participation, transfer, or related learning metrics and then adjust a programme, its content, or its delivery. The evidence supports treating evaluation as a decision input, rather than an end-of-course reporting ritual. Still, an explicit mention can range from routine metric tracking to a larger evaluation responsibility; the binary code does not measure depth, quality, or time spent.

Analysis appears in 89 records. Those descriptions ask practitioners to investigate needs, clarify performance gaps, understand audiences, gather stakeholder requirements, or connect learning to work. This aligns with U.S. public guidance that starts from the performance requirement and the causes of a gap before selecting training. The U.S. Office of Personnel Management presents needs assessment as a mission-linked process for identifying performance requirements, gaps, causes, and development responses. The U.S. Centers for Disease Control and Prevention similarly notes that policy, technology, work environment, or process factors may explain a problem, so training may not be the suitable response in every case. These are attributed public-sector practice sources, not evidence that every private employer follows one prescribed method ([OPM, Planning & Evaluating](https://www.opm.gov/policy-data-oversight/training-and-development/planning-evaluating/); [CDC, Assess Training Needs](https://www.cdc.gov/training-development/php/about/assess-training-needs-conducting-needs-analysis.html)).

Facilitation appears in 78 records. This includes live delivery, workshops, guided practice, coaching of facilitators, and learning discussions. The count is lower than planning, evaluation, and improvement, which is consistent with a labour market containing designers, programme managers, analysts, and learning-technology roles as well as trainers. It would be wrong to infer that the remaining 32 roles never facilitate: the descriptions simply did not provide explicit qualifying evidence under the coding rule.

Learner assessment appears in 47 records, the lowest count. This result needs disciplined interpretation. It may reflect genuinely narrower responsibility, less detailed vacancy wording, or descriptions that refer to evaluation without identifying a learner-level check. The protocol deliberately did not convert generic “evaluation” into “assessment.” That decision prevents a common conceptual error: measuring whether participants learned is different from evaluating whether a programme produced useful workplace results.

### 3.2 First-party and board strata

The accepted corpus preserves separate counts by source stratum.

| Work area | First-party, n=68 | Board fallback, n=42 |
|---|---:|---:|
| Analysis | 48 | 41 |
| Learning plans | 61 | 42 |
| Facilitation | 57 | 21 |
| Learner assessment | 32 | 15 |
| Programme evaluation | 58 | 35 |
| Improvement | 56 | 35 |

These strata are reported for transparency, not as comparable probability samples. The first-party records include a large set of employer-controlled pages collected across several searches. The board group reflects the sources and editorial audience of the accessible boards. Different wording, role mix, employers, search routes, and current-status assurance all affect the counts. For example, all 42 board records mention learning plans, while 61 of 68 first-party records do. That difference does not establish that board-sourced employers value planning more. It shows why source composition must remain visible when interpreting a purposive corpus.

The board fallback was still useful. It supplied readable duties where a current first-party page could not always be inspected, and it broadened the observable work. Its inclusion also introduces a limit: employer-controlled currency and exact wording are less certain. The report therefore uses combined counts only as descriptions of accepted records and avoids a claim about the U.S. labour market as a whole.

### 3.3 A connected workplace workflow

The strongest substantive conclusion is structural rather than statistical. The six categories form a sequence of professional decisions, although real projects may loop or begin in different places.

First, practitioners clarify the performance situation. What work result needs to change? Who performs the work? What is already happening? Are knowledge and skill the likely constraints, or are process, tools, incentives, capacity, access, and management support more plausible causes?

Second, they specify a learning plan. A defensible plan identifies the target group, objectives, practice conditions, resources, delivery arrangements, constraints, ownership, and evidence to collect. It should explain why its chosen response fits the diagnosed problem.

Third, they enable practice and facilitation. Some roles deliver directly; others create facilitator materials, coach subject-matter experts, or design experiences that managers and peers help sustain. Delivery is therefore both a live event and an operating arrangement.

Fourth, they assess learners where evidence is needed. A recall question can test knowledge; a worked scenario can test judgement; an observed demonstration can test performance. The assessment should match the claimed capability. Completion data alone usually cannot show competence.

Fifth, they evaluate the programme in relation to its intended result. Useful evaluation begins before delivery, because baseline, comparison, timing, data ownership, and acceptable evidence need early decisions. Participation and satisfaction can describe experience, while transfer and workplace indicators answer different questions.

Sixth, they improve the response. Improvement may mean revising an example, changing practice time, supporting managers, removing an access barrier, updating materials, or recommending a non-training change. The evidence base supports iteration, but it does not validate one universal branded model.

This workflow is also consistent with the U.S. Occupational Information Network profile for Training and Development Specialists, which lists needs inquiry, programme development, varied delivery methods, programme records and effectiveness, and alternative approaches when improvement is insufficient. Occupational Information Network is used here as an occupational task cross-check, not a vacancy-frequency source ([Occupational Information Network OnLine, 13-1151.00](https://www.onetonline.org/link/summary/13-1151.00)).

### 3.4 Learner assessment and programme evaluation

The 47 learner-assessment mentions and 93 programme-evaluation mentions demonstrate why the concepts should not be merged. Suppose an organisation introduces a new case-handling process. A learner assessment might ask an employee to classify a case, choose the next action, explain a decision, or complete a simulated task. Its unit of analysis is the learner's evidence against a capability criterion.

A programme evaluation asks a broader question: did the intervention work well enough for its purpose, for whom, under what conditions, and at what cost or trade-off? It might combine participation, assessment, transfer checks, quality indicators, supervisor observations, operational data, and learner access information. A good learner assessment may contribute evidence, but it does not establish that the programme caused a workplace change. Conversely, an operational metric may improve while learner performance remains untested or while other changes explain the result.

For practice, this means L&D teams should define claims before choosing measures. “Participants attended,” “participants can perform the task in a simulation,” “participants use the behaviour at work,” and “the business indicator improved because of the programme” are progressively different claims. Each needs different evidence. Before-and-after movement alone is particularly vulnerable to competing explanations such as seasonality, new tools, staffing changes, manager attention, or policy updates.

### 3.5 Audience and seniority

The sampled roles range across specialist, coordinator, trainer, designer, programme manager, director, and mixed leadership or technical positions. The evidence therefore supports three audiences for professional learning.

Specialists and early-career practitioners need a clear operating sequence and tangible outputs: a problem statement, stakeholder map, evidence plan, learning objective, practice design, facilitator guide, assessment, evaluation brief, and improvement decision. Senior practitioners need the same foundations plus portfolio choices, governance, resource trade-offs, stakeholder negotiation, and the discipline to reject training when the evidence points elsewhere. Mixed audiences benefit when shared tools clarify hand-offs: a manager can state the performance outcome, a specialist can design evidence and practice, and a subject expert can verify task accuracy.

This is an educational design implication, not a claim that one programme fits every industry or seniority level without adaptation. Regulated or technical roles need additional subject expertise, authorised procedures, and sometimes credentials. Vacancy evidence should shape realistic practice, not promise eligibility for every sampled job.

## 4. National occupational context

The U.S. Bureau of Labor Statistics reported 472,500 training and development specialist jobs in 2025. Its Occupational Outlook Handbook, updated 27 August 2026, projects 11% employment growth from 2025 to 2035 and about 46,000 openings each year on average. The annual openings include replacement needs as people transfer occupations or leave the labour force; they are not 46,000 newly created jobs per year. BLS also reports employment across several named industries, including healthcare and social assistance, professional, scientific and technical services, educational services, finance and insurance, and administrative and support services ([BLS, Training and Development Specialists](https://www.bls.gov/ooh/business-and-financial/training-and-development-specialists.htm)).

These national figures provide occupational scale and outlook. They should not be combined mathematically with the 110-record vacancy sample. The BLS profile uses a defined national occupation and formal projection methods; the present study is a purposive snapshot of publicly readable advertisements. The cross-industry BLS distribution supports the limited statement that the occupation is employed in several major industries. It does not support a claim that this study represents all industries or that every sector uses the same L&D workflow.

Named employer announcements provide additional context about workplace learning choices. Walmart has described an employee technician pathway combining hands-on and classroom learning and Academy development in service, leadership, and technical areas ([Walmart, 5 June 2025](https://corporate.walmart.com/news/2025/06/05/walmart-pilots-ai-interview-coach-and-accelerates-associate-to-technician-program-to-fuel-careers)). Amazon reports that 425,000 U.S. employees were among more than 700,000 global participants supported by its upskilling commitment and describes learning scheduled around work ([Amazon, Future Ready 2030](https://www.aboutamazon.com/news/workplace/amazon-future-of-work-skills-jobs-training)). These are employer self-reports and announcements. They illustrate investment, scheduling, pathways, and access; they do not prove causal returns, realised future spending, national prevalence, or demand for a particular certificate.

European evidence is comparative only. Cedefop's 2023 profile used a broad teaching-professional group and historical 2021 and 2022-2035 evidence ([Cedefop 2023 profile](https://www.cedefop.europa.eu/en/data-insights/teaching-professionals-skills-opportunities-and-challenges-2023-update)). Its 2025 release identified teaching professionals among broad groups expected to have high openings ([Cedefop 2025 forecast release](https://www.cedefop.europa.eu/en/news/cedefop-skills-forecast-2035-twin-transition-and-demographic-challenge-drive-demand-high-level)). Cedefop issued an updated Skills Forecast release in July 2026, but current teaching-specific figures were not extracted for this study ([Cedefop July 2026 update](https://www.cedefop.europa.eu/en/news/cedefop-updates-skills-forecast-new-data-guide-europes-skills-and-labour-market-policy-2035)). The older numbers therefore cannot be presented as current 2026 estimates. The European group has no direct equivalence to U.S. SOC 13-1151 and cannot validate U.S. corporate L&D prevalence. Public ATD figures were excluded from quantitative conclusions because respondent geography could not be established.

## 5. Implications for practice

For employers, the findings favour commissioning language that begins with a work result rather than a training format. A request such as “build a two-hour workshop” closes key decisions too early. A stronger request states the observed gap, affected group, operational context, available evidence, constraints, and decision deadline. L&D can then recommend learning, performance support, process change, manager action, or a combination.

For practitioners, planning should connect objectives, practice, assessment, transfer, and evaluation. If the objective claims judgement, the practice and assessment should require judgement. If the intended result depends on manager reinforcement or tool access, the plan should specify those conditions. Evaluation should be proportionate to the decision: a pilot investment may require stronger evidence than a routine content update, while a low-risk change may need a concise transfer check.

For professional education, the sample supports applied tasks spanning all six areas. Learners should practise diagnosing a situation, writing a learning brief, designing active practice, facilitating a decision-rich session, creating an appropriate learner assessment, defining an evaluation plan, and making an improvement recommendation. Cases should include senior and specialist perspectives without pretending that a general course supplies regulated technical competence.

Artificial intelligence can assist with bounded preparation: organising stakeholder notes, producing draft variants from supplied facts, identifying ambiguities, or critiquing alignment. Human practitioners remain responsible for source accuracy, privacy, bias, accessibility, operational fit, assessment consequences, and final decisions. No fabricated AI output is evidence of a workplace result.

## 6. Limitations

The sample is purposive, nonprobability based, and limited to publicly accessible English-language material observed on one date. It overrepresents searchable employers and accessible ATS or board sources. Source and employer concentration, differing board practices, and missing exact posting dates reduce representativeness and freshness assurance. Public status can change after retrieval.

Binary coding records explicit mention, not the depth, quality, importance, or share of work. Non-mention is not absence. Roles and sectors are heterogeneous, and titles do not provide a consistent seniority scale. The source strata have different compositions and are not designed for causal or inferential comparison.

The targeted independent audit did not duplicate-code the entire corpus. No inter-rater score is available. Context sources have their own populations and dates: BLS is a national occupational forecast, Occupational Information Network an occupational task profile, OPM and CDC public-sector practice guidance, named employer sources self-reports, and Cedefop European comparative evidence. None establishes a causal effect of training or a return from the proposed education.

The study does not estimate all U.S. vacancies, skill prevalence in the occupation, salaries, hiring probability, credential recognition, learner demand, sales, promotion, or business return. It does not claim that a course creates job readiness for senior, specialist, regulated, or technical posts.

## 7. Reproducibility and data access

The published archive package comprises eight files: `learning-development-us-employer-demand-practice-2026.pdf`; `accepted-vacancies-v1.csv`, containing the 110 accepted records; `dispositions-v1.csv`, containing all 114 candidate dispositions; `coding-summary.json`; `vacancy-coding-evidence.json`, containing sanitized record-level codes, coding bases, status evidence, and limitations; `context-sources.json`, containing the principal and comparative bibliography; `methods-v1.md`; and `data-dictionary-v2.md`. The accepted CSV has SHA-256 `11b4416d3cda94fff3e48bd90f783768232ff2d64f6442e05cc5469457bb2634`. Together, these files expose the selection flow, analysis rules, derived counts, and source boundaries without redistributing full vacancy descriptions or private production material.

Reproduction means applying the published inclusion, deduplication, and coding rules to the preserved evidence. It does not mean assuming that a source URL remains unchanged. The observation date matters because vacancies can open, close, move, or change after collection.

The observation date is 4 September 2026. Targeted independent review occurred on 4 and 5 September, and publication review was completed on 5 September 2026. Later replication may produce different results because vacancies open, close, move, or change.

## 8. Conclusion

The accepted sample describes L&D as a connected workplace practice. Planning, analysis, facilitation, learner assessment, programme evaluation, and improvement are related but distinct decisions. Planning appeared most often, while learner assessment appeared least often under a deliberately strict coding rule. Evaluation and improvement were prominent, underscoring that delivery is only part of the job.

The practical message is to move from a requested learning event to a defensible result chain: diagnose the work problem, select a suitable response, design practice and support, gather evidence at learner and programme levels, and improve the system. That sequence reflects the sampled vacancies and relevant U.S. occupational and public guidance without treating a one-day sample as the national labour market.

## Rights statement

Copyright © 2026 MTF Institute. Except for third-party material, this report and its original analysis are licensed under the Creative Commons Attribution 4.0 International licence (CC BY 4.0). Third-party names and trademarks belong to their respective owners, and linked source pages remain subject to their publishers' terms and rights. The CC BY 4.0 licence does not relicense linked pages, employer marks, or other third-party material. The report reproduces no proprietary model, paid table, assessment instrument, vacancy description, logo, or third-party checklist. No source cited here endorses MTF Institute or any related programme. This statement describes reuse terms and is not a legal guarantee.

## Continue learning

Apply the evidence from this report through MTF Institute's [Professional Certificate in Learning and Development](https://mtfinstitute.com/programs/learning-and-development/#enroll). The programme turns the identified capabilities into structured theory, guided AI practice and reusable workplace artifacts.



## Citation

When citing or summarizing this material, link to the canonical HTML page: https://mtfinstitute.com/insights/learning-development-us-employer-demand-practice-2026/