LEARNING & DEVELOPMENT
Learning and development, built as a supply chain: forecast capability demand, prove the return in cost-per-hire terms, and measure what a CFO will accept.
The most expensive capability gap I have watched open up was not a skills gap. It was an experience gap. The people had the credentials. They had never done the work under pressure.
The pattern repeats. The budget goes out, the completion dashboards look healthy, and the capability the business needs is still missing when the project starts.
Training happened. Capability didn’t.
Deloitte put a number on it in 2025: 66% of managers and executives said recent hires were not fully prepared, and experience was the failing they named most often.
Source: Deloitte, 2025 Global Human Capital Trends
Translation: the gap most organisations are failing to close is not a hiring problem. It is a design problem sitting inside learning and development.
$500a day
The operational drain of one unfilled senior engineering role, measured across the roles I have owned. Sixty days of vacancy is $30,000 of lost productivity, and that is the bill for capability nobody built.
Most organisations run learning and development as a catalogue. I run it as inventory.
Inventory means I forecast the capability the business will need, build it before the requisition exists, and report it in the same currency the finance team already uses for hiring. Catalogues get measured on what was delivered. Inventory gets measured on what was ready.
Here’s what this guide contains:
- The seven-step strategy build, from forecasting capability demand to running it as a quarterly loop
- The metric set a CFO will sign off on, and the three metrics to stop reporting upward
- A delivery-method decision table with the honest failure mode named for every method
- The learning and development and talent acquisition budget crossover, with real cost-per-hire arithmetic
The Lab approach: every number on this page comes from a hiring profit and loss line I have owned or a named, dated source, never a vendor infographic. We publish field-tested blueprints across the Lab, and this one sits inside the wider talent management function alongside recruitment, onboarding, and performance.
What is Learning and Development?
Learning and development (L&D) is the organisational function responsible for building the skills, knowledge, and behaviours a business needs to execute its strategy, through structured training, on-the-job experience, coaching, and career development. For an HR leader, it owns three things: capability supply, compliance assurance, and internal mobility readiness.
Those three are worth stating out loud, because most L&D functions inherit none of them in writing. Capability supply means the skills the business needs are present before the work arrives. Compliance assurance means mandatory training is complete, current, and auditable.
Mobility readiness means people are developed enough that an internal move is a real option rather than a theoretical one. Each of the three is measurable. None of them is “employee growth”.
Here’s the difference:
The old way defines L&D by its activities: it delivers training programmes, runs the induction, manages the platform. The Lab Way defines it by its output: the organisation has the capability it needs, when it needs it. Activity-based definitions are exactly why L&D gets measured on completion rates.
That definition also draws a boundary. L&D is not accountable for performance failures caused by role design, unclear expectations, or an untrained manager. Those sit elsewhere in the talent management system, and they need different fixes.
⚠ WATCH OUT
Common mistake: accepting an undefined remit in January and inheriting the blame in December for outcomes the function never controlled. Write the three accountabilities into the charter, then say them out loud when scope creep arrives.
◆ FROM THE LAB
The Sofia test: I have watched a team commission a leadership programme to fix what was actually a role-design problem. The programme ran, the feedback scores came back fine, and the same escalations returned within a quarter.
Capability problems are mostly operating-model failures, not training failures. I apply the same test to development that I apply to retention: before I design an intervention, I check whether the system around the person is the thing that is broken.
Learning and Development vs. Training vs. Talent Development
Five terms get used interchangeably in most organisations. They are not interchangeable. They map to different budget lines, different owners, and different success measures, which is why the person who can articulate the difference is usually the person who gets funded.
Here’s what actually separates them:
| Term | What it covers | Time horizon | Typical owner | Success measure |
|---|---|---|---|---|
| Training | Closing a known skill gap in a known role | Days to weeks | Line manager, with L&D delivering | Competence at the task, assessed after |
| Learning and development | Building capability for roles and work that do not exist yet | Quarters to years | L&D, accountable to the business plan | Capability coverage against forecast demand |
| Talent development | Growing named individuals toward defined future roles | One to three years | HRBP, with L&D and the business | Readiness coverage for critical roles |
| Enablement | Equipping one function to run one process well | Weeks to months | The function itself, often sales or service | That function’s own output metric |
| Organisational development | Changing structure, culture, and ways of working | Years | HR leadership, with the executive team | Structural and behavioural change at system level |
The sharpest distinction is time and certainty. Training closes a known gap in a known role. Learning and development builds capability for roles and work that do not exist yet. Training is an activity. Learning and development is a function that uses training as one of several methods.
◆ PRO TIP
The catch: these boundaries are cleaner on paper than in an org chart. In a smaller organisation one person owns all five, and the distinction still matters, because the budget conversation and the success measure both change depending on which one you are actually funding.
The practical test is the budget line. If you cannot name which of the five you are asking money for, the request gets approved as training and measured as training. The same discipline applies one level down, where structuring development for an individual employee forces the same choice between closing a gap today and building for a role that does not exist yet.
Why Learning and Development Fails in Most Organisations
The function has a reputation problem, and it earned it. Search for corporate learning and development and the loudest voices are practitioners asking whether any of it works. Three failures produce that reputation, and all three are structural rather than personal.
The completion-rate trap
When completion rate is the number that gets reported, the rational response is to raise it. Shorter modules, easier assessments, mandatory assignment. All three lift the percentage and lower the capability.
Nobody in that chain is acting in bad faith. They are optimising the metric they were handed.
⚠ WATCH OUT
Common mistake: reporting a dashboard that would look identical whether or not anyone’s performance changed. If that is true of yours, it is measuring attendance.
◆ FROM THE LAB
The Sofia test: I ran this exact trap in hiring for years. I counted interviews, screens, and submissions, all of which climbed while nothing actually improved.
Then I switched to Time-to-Productivity, and the number stopped flattering me: 28 days for internal moves against 67 for external hires. Learning and development needs the same switch.
The scale of it is not anecdotal. Only 36% of organisations run career development programmes that produce business results, 31% have programmes with limited adoption, and 33% have nothing running at all. Two-thirds are funding development that nobody uses or that does not exist.
The untrained manager problem
Whatever the L&D team designs, the manager decides whether it gets used. The manager approves the time, sets the stretch assignment, gives the feedback, and signals whether development is real or performative.
If managers have not been trained to hold a development conversation, the programme dies on contact with the org chart.
The old way
- Launch the programme to employees
- Send managers a comms email
- Chase completion
- Report adoption
The Lab Way
- Train managers on development conversations first
- Hand them a two-page framework, not a workshop
- Give them their own team’s capability gaps
- Make conversation quality a manager metric
I hold the same position on retention: those are systems failures, not compensation failures, and the systems that fail most are manager capability, role clarity, and internal mobility. The same three explain why development underdelivers, which is why a development programme built around manager conversations outperforms one built around content.
Fewer than half of managers worldwide have received any management training. The distribution network for every programme you launch is untrained by default.
Buying the platform before defining the problem
The sequence is predictable. A capability problem surfaces, a demo gets booked, the platform is purchased, the content library is switched on, adoption stalls, and L&D owns the failure.
The question that prevents this:
Which specific capability gap does this close, and how will we know it closed? If the answer has to come from the vendor, do not buy yet.
◆ PRO TIP
The catch: platforms are genuinely necessary at scale, for compliance auditability and tracking no spreadsheet survives. The failure is sequence, not tooling. The most capable systems I have worked with were platforms rather than plug-and-play tools, and they punish lean teams that lack change-management discipline.
The Business Case: What Learning and Development Returns
Every competing page has a benefits section, and every one of them is a list. Eight one-line claims, no mechanism, no numbers. Here are three returns instead, each with a mechanism and a figure, written for the person who has to defend the line item.
The three returns that survive scrutiny
Return one: development access moves engagement, and engagement moves retention.
State the chain rather than asserting the conclusion. Access to development predicts engagement, engagement predicts discretionary effort and intent to stay, and intent to stay reduces replacement cost.
Access is rarer than most leaders assume. Only 32% of U.S. employees say they have had opportunities to learn and grow at work in the past year. Two in three people are not getting the thing that moves the first link in that chain.
Source: Gallup, Addressing the Barriers Blocking Employee Development
◆ PRO TIP
Real talk: this chain is correlational at population level. Do not walk into a finance review calling it guaranteed causal ROI. Present it as risk reduction against a replacement cost the business already books.
Return two: internal mobility is the cheapest hiring channel you own.
A developed internal candidate avoids the agency fee, avoids most of the vacancy cost, and reaches productivity in roughly half the time of an external hire. That return belongs to learning and development, not to recruitment. The internal candidate was only available because someone developed them eighteen months earlier, which is what the development strategies that produce a ready internal bench are actually buying you.
◆ FROM THE LAB
The Sofia test: at one point when I owned end-to-end hiring for an engineering business, external cost was climbing and ramp time was the problem nobody had costed. Senior roles took 70 to 80 days to fill, and agency fees ran 18 to 22% of CTC.
We introduced a First Look policy: a 48-hour internal posting window before anything went external, with recruiters actively tapping internal candidates instead of waiting for applications. Small process change, uncomfortable implication. It forced hiring managers to look properly at people they had already written off as not ready.
Within the year, 23% of hires were internal moves and agency reliance fell by 60 to 70%. The number that changed my mind was Time-to-Productivity.
Your best candidates are sometimes already on payroll. They are only ready if somebody built them first.
| Metric | Before | After |
|---|---|---|
| Cost per hire, senior engineering | ₹6.5–7.5L | ₹3–3.8L |
| Time to fill | 70–80 days | 45–50 days |
| Time-to-Productivity | 67 days, external hire | 28 days, internal move |
A 45 to 55% reduction in cost per hire is a learning and development return. It just gets booked to recruitment.
Return three: capability built early removes the vacancy bill.
The $500 a day from the opening is the same money viewed from the other end. Building the successor in advance does not shorten the hiring process. It removes the need to run one.
What the business case cannot claim
Three things this case cannot honestly claim. Direct causal attribution of revenue to a training programme, because the confounds are unmanageable. A universal ROI multiple, because the figures in circulation are old, vendor-amplified, or both.
And short-horizon returns. Capability built this quarter surfaces two to four quarters out, which is exactly why learning and development loses budget arguments to initiatives that report monthly.
◆ PRO TIP
The honest downside: you will not win this on enthusiasm. Agree the attribution model with finance before the programme starts, and name these three limits yourself before a CFO names them for you.
The Core Components of a Learning and Development Function
Most L&D functions can name their components. Very few can say who owns each one and how each is measured, which is why the taxonomy never turns into an operating model. Six components, each with a named owner and a single measure.
1
Onboarding and role readiness
Getting a new or newly moved person to full contribution in the first 90 days, which is where structured employee onboarding either delivers or quietly fails. Not a completed induction checklist.
Who owns it:
The hiring manager, with L&D designing.
How it’s measured:
Time-to-Productivity.
2
Compliance and mandatory training
The training the organisation is legally or contractually obliged to deliver, kept current and auditable. This is the one component where completion rate is the correct metric.
Who owns it:
L&D, with legal setting the requirement.
How it’s measured:
Completion against the obligation, with evidence that survives an audit.
3
Technical and functional upskilling
Deepening capability inside the role people already hold. Method choice matters more than volume here, and the training methods that fit each gap type decide whether any of it lands.
Who owns it:
L&D, with the function’s technical leadership.
How it’s measured:
Capability coverage against forecast demand.
4
Leadership and manager development
Equipping managers to run their teams, and to hold the development conversations every other component on this list depends on.
Who owns it:
HR leadership, with L&D delivering.
How it’s measured:
Development conversation quality, assessed inside the manager’s own review.
5
Reskilling and redeployment
Moving people out of declining roles and into growing ones, before the decline turns into a redundancy conversation.
Who owns it:
The HR Business Partner, with L&D delivering.
How it’s measured:
Redeployment rate against attrition avoided.
6
Career pathing and succession
Visible routes to the next role, and named, ready internal candidates for the roles the business cannot afford to leave open.
Corporate Learning and Development at Scale
Corporate learning and development is not the same function with more people in it. Four things break at scale, and they break in ways that make it look like the programme failed when the context did.
What changes when L&D goes multi-site
- Consistency against relevance. A single global curriculum is consistent and irrelevant. Hold a mandatory core plus a local layer, at a stated ratio.
- Language against localisation. Translation is not localisation. The examples have to change, not only the words.
- Compliance divergence. Requirements differ by jurisdiction, so the tracking has to be jurisdiction-aware or the audit fails.
- Manager variance. At scale, manager capability varies more than employee capability, so one programme produces very different outcomes by site.
◆ FROM THE LAB
The Sofia test: I have run hiring and capability programmes across multiple sites and countries inside an engineering business. Nobody warns you that the same material lands completely differently in two plants forty miles apart, and the variable is almost always site leadership rather than content.
⚠ WATCH OUT
Warning: before concluding a programme failed, compare outcomes by site. If three sites moved and two did not, you have a manager problem or a context problem, and rebuilding the content will fix neither.
One caveat belongs in the plan itself. Mandatory training requirements and record-keeping obligations differ by country, and in some cases by state or province, so a single global compliance approach is not sufficient.
Running a capability build like a campaign
A large capability build is a logistics project that people keep planning as a content project. The constraints match a hiring ramp-up exactly: fixed deadline, limited facilitator capacity, high drop-off risk.
1
Map backward from the business date
Fix the date capability must be present, then work backwards to cohort start dates. Build from the deadline, not the calendar.
2
Forecast bandwidth as a constraint
Facilitator hours and manager release time are hard limits. Forecast them as constraints, never as assumptions.
3
Batch, and score the same way
Run cohorts in batches against structured scorecards, so quality is comparable across sites rather than anecdotal.
4
Stand up against a live tracker
A short weekly stand-up against a real-time tracker. Steering beats reporting, and only if the tracker exists first.
The mechanics hold whether the cohort is being hired or being developed, which is why running employee training programmes at volume looks far more like a hiring campaign than a curriculum design exercise.
◆ FROM THE LAB
The Sofia test: at one point I owned a ramp-up of roughly 350 to 500 people across customer support, operations analysts, junior engineers, and sales support. The window was 8 to 12 weeks against an aggressive business start date, with limited interviewer capacity and a real risk of people dropping out faster than we could replace them.
We planned it as logistics. Weekly targets were mapped backward from the business start date, recruiter bandwidth and interviewer availability were forecast as hard constraints, and roles were clustered so the pool widened without the bar dropping.
Then we industrialised delivery. Standardised pre-assessment criteria, batch interviews with structured scorecards, daily stand-ups against a real-time tracker, weekly war-room calls with hiring managers, and SLA-based vendor governance. Every candidate had one point of contact per batch and a clear timeline, which is what actually cut the drop-off.
We delivered 420 hires in 10 weeks. Average time to fill fell from 32 days to roughly 18 to 20, offer-to-join held above 90%, and 90-day retention landed at 85 to 88%. Recruiter productivity improved by about 35%.
The transfer is direct. The same capacity forecasting, batching, live tracking, and war-room cadence are what make a 400-person capability build work, and dropping any one of them is where these programmes quietly fail.
Employee Learning and Development Across the Lifecycle
Employee learning and development is usually front-loaded into onboarding, and then it stops. The people holding the most organisational knowledge end up receiving the least investment, which is exactly where quiet capability loss happens. Four stages, each with its own intervention and a single measure.
1
First 90 days: role readiness
The need is real contribution, not a finished induction. Give the person a named first deliverable inside 30 days, a manager check-in cadence that survives a busy quarter, and direct access to the two or three people whose knowledge the role depends on. This is where structured employee onboarding either delivers or quietly fails.
The measure:
Time-to-Productivity.
2
Months 4 to 18: depth and autonomy
The need is delivering core work without escalating it. Hand over progressively harder cases, give feedback tied to specific work rather than to a review cycle, and agree one capability target with the manager. The failure here is subtle: the person becomes competent and invisible, and development stops because nothing is going wrong.
The measure:
Independent delivery of core work, without escalation.
3
Year 2 onward: breadth and mobility
The need is exposure beyond the current role, before the next role opens. Provide cross-team projects, a stretch assignment with real consequences, and a visible route to the next role instead of a vague conversation about growth, which is what a written career development plan is actually for.
The measure:
Internal mobility rate, plus readiness coverage for the next role.
4
The plateau: re-engagement or redeployment
The need belongs to the strong performer who is neither leaving nor growing. Run a deliberate re-engagement rhythm rather than waiting for a trigger, open the redeployment conversation before the role declines, and be honest about what is genuinely available. No competitor page covers this stage, and it is where most organisations lose their deepest capability.
The measure:
Re-engagement or redeployment rate.
⚠ WATCH OUT
Warning: the plateau is invisible in every dashboard you own. Nobody is failing, nobody is complaining, and the engagement score is unremarkable. Then a person with eight years of context resigns, and the capability walks out inside a week.
◆ FROM THE LAB
The Sofia test: I built a Silver Medalist Re-Engagement Loop for strong candidates we did not hire, contacting them deliberately at 3, 6, 9, and 12 months rather than only when a role opened. Four hires came out of it across two years, and one of them became a team lead.
The same rhythm works on plateaued internal talent, and almost nobody runs it. We re-engage external candidates on a schedule and internal ones by accident.
Keep this stage framed on development and redeployment. In several jurisdictions, conversations about declining roles move quickly into redundancy territory, and that is a different process carrying different obligations.
How to Measure Learning and Development (metrics a CFO will accept)
Every competitor names Kirkpatrick’s four levels and stops there. None of them says the quiet part: levels 3 and 4 are rarely measured, because they need data nobody captured and an attribution model nobody agreed.
Why Kirkpatrick stalls at level 2
The four levels are reaction, learning, behaviour, and results. Levels 1 and 2 are cheap to capture, so they get captured. Level 3 needs observation over time against a baseline that was never taken, and level 4 needs an attribution model agreed before the programme ran.
So the field reports levels 1 and 2 and calls it evaluation. That is not dishonesty. It is the only thing the available data supports.
◆ PRO TIP
Real talk: you will not reach level 3 on a programme that has already been delivered. Capture the baseline before the next one starts, and agree the attribution model with finance at design time. Both of those are Step 6, which is precisely why Step 6 sits before delivery.
◆ FROM THE LAB
The Sofia test: I can tell you that competition analysis moved quality of hire by 12%, talent landscape work by 14%, and skill-based pipelines by 30%. Those numbers exist for exactly one reason. The baseline was captured before each change, not reconstructed afterwards.
The metric set that survives a finance review
Eight metrics, two tiers, and a stop-doing list. A twenty-metric dashboard is indistinguishable from no prioritisation, and it belongs inside the wider set of talent metrics rather than in a report of its own.
| Metric | Tier | What it tells you |
|---|---|---|
| Internal fill rate | Board | Whether capability is being built or bought |
| Time-to-Productivity, internal against external | Board | Whether development actually shortens ramp |
| Cost per hire trend | Board | The return, in the currency finance already uses |
| Critical-role readiness coverage | Board | Share of critical roles with a ready-now internal successor |
| Regretted attrition in capability-critical roles | Board | Whether the capability you built is staying |
| Capability coverage against forecast demand | Operational | Whether the plan is closing the gaps it named |
| Development conversation completion by manager | Operational | Whether the distribution network is working |
| Application rate within 90 days, sampled | Operational | Whether learning is reaching real work |
Stop reporting these upward:
- Hours delivered. It measures effort, not outcome.
- Completion rate. Outside compliance, it measures attendance.
- Satisfaction scores. They measure the room, not the capability.
Report activity upward and you train executives to ask about activity.
◆ FROM THE LAB
The Sofia test: the numbers that changed a leadership conversation were cost per hire falling 45 to 55%, time to fill dropping from 70 to 80 days to 45 to 50, and agency reliance cutting by 60 to 70%, which avoided agency hires on four of six senior roles at ₹8 to 10 lakh each, roughly USD 9,500 to 12,000.
All Tier 1. Not one of them is a learning metric.
Reporting cadence and who sees what
Three audiences, three packs, one rule. Operational metrics never travel upward.
| Audience | Cadence | Contents |
|---|---|---|
| L&D and HRBP team | Monthly | Tier 2 operational metrics, for steering |
| HR leadership | Quarterly | Tier 1, plus commentary on what changed and why |
| The board | Annually | Tier 1 trend only, with the cost comparison |
⚠ WATCH OUT
Warning: send operational metrics upward once and you teach executives to ask about completion rates. After that a capability conversation is almost impossible to start, because you supplied the vocabulary yourself.
Write one more line into the reporting standard. Completion records, assessment scores, and manager ratings are employee personal data, so GDPR and equivalent data-minimisation rules apply. Report in aggregate upward, and keep individual-level data where it belongs.
Learning and Development and Talent Acquisition: The Same Budget
This is the section no L&D writer can produce, because it needs hiring cost data. Build and buy draw on the same capability budget, and in most organisations nobody decides between them deliberately. The requisition simply opens, which is why hiring strategy and capability building belong in one conversation.
The build-or-buy decision nobody makes deliberately
Four questions, asked before a requisition opens rather than after it stalls.
- Does anyone internally already hold 70% of this capability?
- What is the realistic time to close the remaining 30%, in weeks, verified against the evidence from Step 2 rather than guessed?
- What is the fully loaded external cost: agency fee, vacancy cost per day, and ramp time to productivity?
- What is the second-order effect? Does building internally open a backfill that is easier to hire, and does hiring externally tell the team that development leads nowhere?
If the internal candidate is at 70% and the gap closes inside the vacancy period, building is cheaper on every axis. The arithmetic is not close.
The old way
- A role opens
- Recruitment sources externally
- L&D is never consulted
- The internal candidate finds out from the job board
The Lab Way
- The four questions run before the requisition
- Both functions see the real cost on each side
- A 48-hour internal window opens first
- The build-or-buy call is documented
◆ FROM THE LAB
The Sofia test: agency fees ran 18 to 22% of CTC, roughly ₹8 to 10 lakh per senior hire, about USD 9,500 to 12,000. We avoided four of six senior agency hires in a single year.
Set that against the $500 a day a vacancy drains and the build case writes itself. The mechanism that forced the question to be asked at all was a First Look policy: a 48-hour internal window before anything went external.
None of this works if the two functions never see each other’s numbers, which is why the recruitment side of the same decision is worth reading alongside this.
◆ FROM THE LAB
The Sofia test: earlier in my career an automation engineering team needed to scale, and the default answer was to open requisitions and wait.
Instead we mapped the market first. Twelve companies across the industrial IoT space: their team structures, who sat where, which groups were being restructured, and which people held the capability we would need. All of it done before the job description existed.
By the time the requisition opened, 37 qualified candidates were identified and warm. Time-to-Offer fell from 68 days to 34.
The transfer to development is exact, and it is the point of this entire section. Mapping talent externally and forecasting capability internally are the same discipline, because both replace reaction with inventory. The only difference is whether the person you built the relationship with is already on your payroll.
The shared metric that forces the conversation
One metric, owned jointly: internal fill rate, the share of open roles filled by internal moves. It works as a forcing function because neither function can hit it alone. Recruitment cannot without L&D having built readiness, and L&D cannot without recruitment genuinely considering internal candidates.
Set it as a joint target in both functions’ objectives and report it in the same pack. Pair it with a quality guard so nobody games it: Time-to-Productivity and 12-month retention for internal moves, reported alongside.
In my case a First Look policy delivered 23% internal hires, with Time-to-Productivity at 28 days against
What a Learning and Development Specialist Actually does
Roughly a third of the search traffic for this term comes from people looking at the role rather than the function. This section is written for whoever is designing or hiring it. One title covers three genuinely different jobs.
The role, and how it differs by organisation size
| Organisation size | What the job actually is | What breaks |
|---|---|---|
| Under 500 people | Generalist. Designing and delivering, mostly onboarding and compliance, one person against everything. | Depth. Nothing gets built properly because everything is being run. |
| 500 to 5,000 | Programme ownership: needs analysis, vendor management, manager enablement. | Reach. One person cannot enable every manager, so coverage becomes a lottery. |
| 5,000 and above | Specialisation into learning design, learning technology, capability strategy, or leadership development, working through business partners. | Distance. The role stops touching learners and starts negotiating with stakeholders. |
What actually matters:
Across all three versions the capabilities rank in one order: business literacy, needs diagnosis, facilitation, learning technology. Most job descriptions rank them in reverse, leading with platform experience and instructional design credentials, then treating commercial understanding as a nice-to-have.
That is how an organisation ends up with someone who can build a course but cannot say which course is worth building, or run the development strategies the role was actually hired to deliver.
I have written and hired against these descriptions across eight years in the function, while running 250-plus hires a year. The ones that worked led with the business, never with the toolset.
Where the role sits and why it matters
The reporting line does more work than the job description. L&D reporting into HR operations gets treated as a service function and measured on delivery. L&D reporting into a Chief People Officer, or into the business, gets treated as capability strategy and measured on outcome.
Same person, same skills, different ceiling. If you want the role to be strategic, the line has to say so before the objectives do.
Objectives cannot outrank structure.
◆ PRO TIP
The catch: you can rewrite an L&D job description in an afternoon. Moving where it reports takes a reorganisation, executive sponsorship, and usually a vacancy to justify it. Do it when the vacancy comes up, because the next window may be years away.
Where Learning and Development is Heading: AI, Skills-first, and The Experience Gap
Trends sections are the third-worst fluff pattern on this topic: three named shifts, no implementation detail, no cost, no risk. Two shifts matter, and each one gets a verdict.
AI in learning and development: what it actually changes
Three uses hold up. Generating and adapting content at speed, so a module gets rewritten for a site rather than translated for it. Semantic search across the content estate you already own, so people find what exists. Recommendation against a stated skill gap, rather than against browsing behaviour.
⚠ WATCH OUT
Warning: three risks are real, and rarely named by the people selling the tools. An automation arms race, where dependency deepens because everyone else is automating. Loss of the human contact that makes development stick. Algorithmic bias, where flawed training data quietly reproduces historical patterns in who gets developed.
◆ FROM THE LAB
The Sofia test: I led AI-driven prompt engineering and digital transformation inside a recruitment function, and the honest summary is that the tooling moved from assistance to execution faster than most policies did.
What I have watched work: sourcing co-pilots running semantic search across an existing database rather than keyword matching, and agentic orchestration handling the full interview lifecycle instead of only scheduling. Skills-first verification does reduce unconscious bias when the signals are specific. Predictive models earn their place on retention risk more than on hiring demand.
My verdict on the general-purpose assistants has not moved. A genuine productivity multiplier, not a recruiter, and by the same logic not a learning and development function.
The operating rule is fixed. AI accelerates thinking, humans own judgment, context, and the final call, and no AI system decides who is eligible for development.
The pressure is externally validated rather than a Lab opinion. Deloitte found learning and development to be the talent process most in need of reinvention because of AI disruption, with 74% saying it is very or critically important to prioritise human capabilities. Reinvention means rethinking where development is aimed, not buying new tools.
Source: Deloitte, 2025 Global Human Capital Trends
One flag to raise with legal early. AI systems influencing development eligibility, promotion readiness, or selection may fall within employment discrimination law, and in the EU within AI Act obligations for high-risk HR systems.
Skills-first, and what it demands that nobody mentions
A skills-first model needs a taxonomy specific enough to act on and stable enough to maintain. That maintenance is a permanent funded job, not a project. Too granular and it collapses under its own weight, too broad and it says nothing.
Start with the capabilities for the five roles that matter most to the business plan. Prove the model there. Expand only when maintenance is genuinely resourced, which for most organisations means never expanding as far as the vendor deck assumed.
◆ PRO TIP
The honest downside: taxonomies do not fail at launch. They fail at month eighteen, when the person who owned it has moved on and nobody has updated a skill definition in two quarters.
When I built skill-based pipelines off market analysis, quality of hire improved by 30%. The caveat matters more than the number. It worked because the capability signals were specific to those roles, not because a generic taxonomy had been adopted.
A taxonomy also records what people can do, not what they have done under pressure. The experience gap from the opening is the harder one, and no skills model closes it alone.
Go Deeper: The Learning and Development Library
Everything above operates at the level of the function. Three areas go deeper than a single page can, and each one has a full guide behind it. Start wherever your current problem actually sits.
Employee development
Employee development strategies
The strategies that produce a bench of ready internal candidates, and the sequencing that decides whether any of it survives the first busy quarter.
Employee training
Employee training methods
Which method suits which gap, what each costs in time and money, and where each one fails. The practical companion to the decision table above.
Career development
Career development plan
The individual-level version of this page: how one person’s route to the next role gets written down, agreed with a manager, and actually followed.