Doing More With 28% Less: How Learning Teams Are Rebuilding Capability
Doing More With 28% Less: How Learning Teams Are Rebuilding Capability
L&D News You Can Understand | ldnucu.com
Listen to the podcast
THE CAPABILITY SQUEEZE
Doing More With 28% Less
On ldnucu.com, the number that frames everything else is this one: median corporate learning and development spending per employee dropped by 28% in 2026 compared to 2025, according to benchmarking data from the Society for Human Resource Management (SHRM). Over that same period, total required employee training hours stayed flat. For learning leaders, the directive is stark: deliver the same volume of compliance adherence, technical onboarding and leadership development with nearly a third less capital per person.
The capability bridge: who pays for the squeeze?
Funding per employee sits 28% lower, yet the span of training hours has to stay the same length. Switch on a lever to see what holds the deck up.
Funding per employee is down 28% and training hours are flat, so the deck sags into a coverage gap. Choose a lever.
| Lever | What it does | Retention risk |
|---|---|---|
| Cut total rewards | Closes the gap on paper | High: employers protected benefits to keep staff |
| Consolidate vendors and subscriptions | Removes third-party content cost | Low, but does not change delivery |
| Modular, peer-led delivery | Changes how each learning hour is built | Low |
That squeeze is colliding with widespread organizational restructuring. The November 2026 Principal Financial Well-Being Index found that nearly 9 in 10 employers (88%) altered their staffing models, skill requirements, core technologies or work designs during 2026. Yet rather than cutting compensation or scaling back health and financial benefits, employers protected total rewards, with more than half naming talent retention as their top HR priority for the coming fiscal year. Gartner research points the same way: employees consistently rank financial security and workplace stability among their primary reasons for staying.
Consider a hypothetical regional hospital system facing a cut to its education budget at the same moment its nursing staff must train on a new fleet of diagnostic machines. Administration cannot cut base pay or shift healthcare costs onto clinicians without losing experienced nurses to a rival facility across town. Because total rewards are locked in to guard retention, the savings have to come from training delivery itself: dropping expensive third-party content subscriptions, consolidating vendor contracts, and moving from external instructor-led workshops to internal, technology-enabled peer training.
That constraint forces L&D teams out of passive content production and into precision capability engineering. When additional authoring headcount is off the table, every learning hour has to be designed for measurable workplace transfer rather than course completion volume.
POLICY AND RISK
How Public Subsidies Absorb Entry-Level Training Risk
A 28% internal training contraction directly changes how much financial risk employers can absorb when hiring entry-level talent. Bringing on an untested worker who needs extensive foundational upskilling becomes a heavy cost center when internal development budgets are tight.
On November 17, 2026, Alan Milburn released a landmark UK review proposing a 1.1 billion pound annual employment support package to unlock 170,000 jobs and apprenticeship opportunities. It works as a risk-mitigation mechanism for employers: a 3,000 pound base cash incentive for hiring young people who have been Not in Education, Employment or Training (NEET) for at least six consecutive months, plus a 2,000 pound barrier premium for candidates such as care leavers or young adults with disabilities, for up to 5,000 pounds per hire.
Stack the incentives for one hire
Tick the conditions that apply to an imagined hire and employer. The stack adds the payments named in the Milburn Review summary.
The hire and the employer
Direct payments for this hire: 0 pounds. Tick a condition to start the stack.
All four payment tiers and the longer benchmarks
- 3,000 pounds base payment: NEET for at least six consecutive months.
- Plus 2,000 pounds barrier premium (up to 5,000 pounds per hire).
- 2,000 pounds small-business premium: fewer than 50 employees.
- 2,500 pounds apprenticeship payment.
- Secondary National Insurance threshold for ages 21 to 24: 5,000 to 10,000 pounds over a five-year glide path, starting in retail and hospitality.
- By 2036: 35% of each youth cohort entering an apprenticeship by age 25. By 2032: an 11% Civil Service entry-level target for ages 16 to 24.
Smaller employers with fewer than 50 staff receive an additional 2,000 pound premium to lower the cost of administrative onboarding, alongside a dedicated 2,500 pound apprenticeship payment. To lower payroll friction further, the review doubles the secondary National Insurance threshold for employees aged 21 to 24 from 5,000 pounds to 10,000 pounds, along a five-year glide path that starts in retail and hospitality.
Administratively, the review creates a unified Youth Opportunity Service as a single front door for career guidance, skills assessments and job matching, paired with a statutory Youth Opportunity Test, binding Opportunity and Support Guarantees, and a conditional Youth Allowance, linked to training participation, that replaces the Universal Credit Health top-up for young applicants. Over the longer term, it targets doubling apprenticeships by 2036 so that 35% of each youth cohort enters an apprenticeship by age 25 (roughly 270,000 under-25 starts annually), redirects 100% of the renamed Growth and Skills Levy to under-25s, and sets an 11% Civil Service entry-level hiring target for ages 16 to 24 by 2032.
DURABLE SKILLS
Why Seven of the Top Ten Skills Are Human
Under tight budgets, you might expect companies to pour the remaining training funds into software coding. Degreed pathway analysis published in November 2026 shows the opposite: 7 of the top 10 priority skills for 2026 are human-centric. Leadership ranks first, followed by Communication, Project Management, Problem Solving and Customer Service in fifth, while Excel (sixth), Data Analytics (seventh) and Python (eighth) sit behind those human capabilities, before Adaptability and Stakeholder Management complete the list.
The rationale comes down to competency half-life. Specialized software and AI tool skills now carry an estimated two-year shelf life before interfaces shift or models automate syntax, whereas human adaptability compounds over decades. World Economic Forum data estimates that tasks requiring empathy, high-stakes leadership, nuanced communication and creative strategy hold only a 13% potential for AI automation.
Which skills outlast the tools?
Three views of the same argument: the ranking, the shelf life and the automation exposure.
| Rank | Skill | Type |
|---|---|---|
| 1 | Leadership | Human |
| 2 | Communication | Human |
| 3 | Project Management | Human |
| 4 | Problem Solving | Human |
| 5 | Customer Service | Human |
| 6 | Excel | Technical |
| 7 | Data Analytics | Technical |
| 8 | Python | Technical |
| 9 | Adaptability | Human |
| 10 | Stakeholder Management | Human |
Seven of the ten are human-centric. Pick a highlight to see how they sit in the ranking.
When interfaces shift or models automate syntax, a tool skill resets. Adaptability and judgment carry forward into the next tool.
World Economic Forum estimate: only 13% of tasks that involve empathy, high-stakes leadership, nuanced communication and creative strategy have potential for direct AI automation.
Choose a view to compare what organizations aim for with what the strongest performers do.
McKinsey's 2025 State of AI report reinforces the divide: while 80% of organizations set cost efficiency as a primary AI objective, top-performing organizations deploy AI to drive growth and business model innovation. Degreed founder David Blake warns that relying on generic large language models to formulate learning strategy yields generic outcomes. He advocates bi-directional mentorship, pairing senior leaders' deep business context with the digital fluency of early-career staff, so that adoption stays grounded in institutional knowledge rather than novelty.
As AI hype cools, human judgment is emerging as the primary differentiator. An enterprise AI tool can draft a technically flawless project proposal that is politically tone-deaf. It takes a professional with stakeholder management and diplomacy to read the room, anticipate executive pushback and intervene before the document goes out.
PRACTICE SCIENCE
Designing AI Roleplay That Survives Monday Morning
Reading a PDF about empathy does not build empathetic skill under pressure. A meta-analysis led by Chernikova et al. (2020), synthesizing 145 primary studies across professional education, found that simulation-based learning yields a large effect size of Hedges' g = 0.85 on complex skills, moving an average learner from the 50th percentile of performance to roughly the 80th compared with conventional passive instruction.
Feedback inside simulations, however, is delicate. Foundational research by Kluger and DeNisi (1996) across 607 feedback interventions found that more than one-third actually decreased performance, by triggering ego-defensiveness or delivering vague personal judgments. Building on this, Hattie and Timperley (2007) established that effective feedback must answer three task-focused questions: Where am I going? (Feed Up), How am I going? (Feed Back) and Where to next? (Feed Forward).
Spot the feedback that backfires
HypotheticalThree invented comments to a learner after a negotiation roleplay. Decide whether each one helps or risks ego-defensiveness.
Answer all three to see the pattern.
Instructional designer Brianna Blaney translates this science into a five-step framework for AI roleplay: define a two-reviewer-testable goal that independent evaluators would score identically; write a private persona backstory with unstated constraints; program reasonable resistance tied logically to that backstory; establish observable behavioral rubrics; and rely on practitioner-written scenarios. In a hypothetical commercial negotiation, giving an AI procurement manager a secret CFO mandate to cut software spend by 10% forces the learner to use diagnostic questioning rather than reciting product features.
Build the roleplay one step at a time
Add each of the five steps to the structure and see how an imagined procurement scenario changes.
Step 0 of 5
Start with the first step.
- Define a two-reviewer-testable goal.
- Write a private persona backstory with unstated constraints.
- Program reasonable resistance tied to that backstory.
- Establish observable behavioral rubrics.
- Rely on practitioner-written scenarios.
Even well-engineered simulations fail if the behaviors vanish on Monday morning. Research by Suzie Bishop at the Center for Leadership Studies describes this as the Monday morning problem: legacy workplace habits stay faster, safer and better rewarded than the new ones. Grounding transfer requires deliberate practice goals in the tradition of Ericsson, such as assigning one specific recurring task to a named team member by Friday, backed by structured debriefs with frontline managers and the removal of systemic contradictions that punish minor developmental mistakes.
READINESS GAP
Bridging the Divide Between Executives and Frontlines
That same pattern of systemic contradiction is stalling enterprise AI rollouts. A November 2026 Hogan Assessments poll of 648 corporate workers found a sharp divide: managers report more positive attitudes toward AI adoption, while non-managerial staff remain skeptical over job security, data privacy and practical usefulness.
Global talent metrics show how shallow readiness remains. ManpowerGroup Talent research indicates that fewer than 1 in 5 enterprises (under 20%) have reached advanced or transformational AI readiness, and only 3% of Chief Human Resources Officers feel highly prepared to manage AI-driven workforce transitions. Coqual Global Lab found that only 2 in 5 leaders (40%) hold an expert understanding of their organization's strategic priorities, and just 35% strongly understand how technical transformation ties to frontline execution.
Why the boardroom and the frontline disagree
Two views: how thin AI readiness really is, and three ways a learning team can answer change fatigue.
The Hogan Assessments poll of 648 workers adds the human side: managers report more positive attitudes toward AI than non-managerial staff, who worry about job security, data privacy and day-to-day usefulness. The poll is described in direction only, so no bar is drawn for it.
Pick a pillar of the Give Me Something Real framework.
The risk is that executives mistake their own speed boost from drafting emails with a chatbot for enterprise-wide readiness, and issue sweeping mandates without role-specific guidance. To counter change fatigue, learning provider Sponge launched its Give Me Something Real framework in November 2026, pairing Real Evidence for boards with Real Experiences for learners and Real Action for daily roles.
Global analyst Donald H. Taylor reinforces the shift in his 2026 L&D Global Sentiment Survey research. As uncritical AI hype subsides, Taylor notes that L&D's value is moving from rapid content generation toward strategy, human capability and organizational impact. He also flags a persistent disconnect: leaders cite proving business value as their top priority even as their tactical focus on learning analytics declines, and talent teams need clearer boundaries to prevent burnout.
MODULAR ARCHITECTURE
Decoupled Compliance, Agentic Workflows, and External Ecosystems
To survive a 28% contraction without sacrificing rigor, organizations are adopting modular content architectures. Panopto introduced Panopto Develop in November 2026 to capture unscripted expert demonstrations during routine operations and turn them into structured, searchable learning units and assessments, with mandatory subject matter expert review before company-wide distribution. In regulatory compliance, Claro experts Lev Alyshayev and Einav Itamar decouple universal statutory rules, such as OSHA lockout/tagout (LOTO) standards, from site-specific machine isolation steps, using the worked-example effect and SCORM and xAPI exports. Alyshayev warns that generative AI in compliance needs strict human verification, because unchecked models can hallucinate legal clauses, omit exemptions or invent call-count thresholds for phrases such as "all reasonable efforts." Federal OSHA rules still require hands-on observation by a qualified supervisor.
A hypothetical aviation maintenance unit shows the mechanical advantage. Treat the overarching safety rule as a fixed base plate and each aircraft hydraulic-valve procedure as an interchangeable block. When a manufacturer updates a torque specification, designers can regenerate that one sub-module in hours rather than rebuilding the fleet curriculum, and technicians complete it on a mobile device, while in-person supervisor sign-off on the hangar floor stays in place.
Base plate and swappable blocks
HypotheticalAn imagined aviation maintenance unit. Run an update and watch what has to change, and what still needs a person.
Start with the manufacturer update.
Order of events: one sub-module is regenerated, the technician completes it on a mobile device, and a senior supervisor observes the step in person before certification. The universal safety rules are untouched.
Beyond modular content, agentic workflows are automating multi-step learning administration. Josh Bersin's analysis of the Galileo AI agent, integrated across platforms such as Sana and Workday, shows how agents query fragmented internal data, run skill gap analyses, draft tailored learning pathways and update worker profiles without manual administration. At the customer-education level, EffectusLMS showcased adaptive recommendations, an in-course conversational assistant, automated scenario generation and real-time learner sentiment analytics at LearnFlux 2026.
External skilling ecosystems are expanding at the same time. Databricks Canada reported over 125% year-over-year revenue growth on November 17, 2026 and committed to training more than 50,000 Canadians in data and AI over three years, with free software access across more than 70 Canadian colleges and universities, an expanded Databricks Academy curriculum and industry-recognized certifications. It addresses a market where Q2 2026 statistics show only 19.2% of Canadian businesses actively use AI. In the public sector, OpenAI signed a non-binding memorandum of understanding with Chile's Ministry of Science, Technology, Knowledge and Innovation on November 17, 2026, covering civil-servant AI literacy, output verification, sensitive-data protection and train-the-trainer workshops.
Macroeconomically, Research and Markets projects the global EdTech market to grow from $187.01 billion in 2025 to $437.54 billion by 2033, a 10.8% compound annual growth rate driven primarily by enterprise AI integration. Meanwhile, Google and NRG research shows 60% of B2B technology buyers now use AI tools in purchasing research, which makes AI mention share, recommendation rate and citation rate the visibility metrics worth tracking.
The outside ecosystem in two numbers and three metrics
Three ways to measure visibility when 60% of B2B technology buyers use AI tools in research
Choose a metric to see what it counts.
- AI mention share: how often a brand is mentioned.
- Recommendation rate: how often a brand is suggested as a top solution.
- Citation rate: how often an organization's primary domain is linked as a source.
THE PATH FORWARD
Aligning Incentives Before Buying More Tools
Across these 2026 workforce developments, one operational reality stands out: AI has dramatically lowered the marginal cost of generating modular courses, updating localized compliance units and querying internal data across enterprise systems.
Yet durable organizational capability still depends on human-centric judgment, stakeholder diplomacy, scientifically structured simulation and localized managerial reinforcement that software cannot automate.
Monday morning incentive check
Four questions to ask before buying another platform or agent. Answer for your own organization.
Answer all four questions to see what to fix first.
You can automate the delivery of a revised technical procedure, but you cannot automate the psychological safety and workplace culture that encourage a frontline employee to apply it without fear of penalty.
Before investing in another scalable training platform or intelligent agent tomorrow morning, inspect your daily operational incentives to make sure your workplace is not quietly rewarding the exact legacy behaviors you are paying to change.
Watch the podcast on YouTube