Public-Sector Delivery Training Must Move Beyond the Classroom

Thought Leadership

Why governments need live assignments, professional standards and a Global Delivery Network to build delivery capability that lasts.

Only 27 percent of surveyed public servants in Bosnia and Herzegovina, 29 percent in Kosovo, and 43 percent in Ethiopia reported receiving government-funded training during the preceding 12 months.

The problem is not only that too few public servants are being trained. Even when training happens, it is frequently disconnected from what institutions and officials actually need. These findings come from the World Bank’s 2026 Public Workforce Performance and Prosperity report. Its evidence base includes 67 million observations across 151 countries and surveys of more than 200,000 public-sector employees in 30 countries. The report concludes that government training is often not anchored in organisational needs or connected to the competencies required for different jobs.

That is a serious problem when the world depends on approximately 400 million public-sector workers to administer public resources, regulate markets, deliver essential services and implement government priorities. Training public servants is therefore not a peripheral human-resources activity. It is part of the operating infrastructure of the state.

Yet much of what is called capacity building still measures attendance, hours completed and participant satisfaction. Those measures tell us whether a programme took place. They do not tell us whether government became more capable of delivering.

The data shows an access problem and a relevance problem

The World Bank’s public-workforce surveys reveal significant differences in access to training.

Surveyed workforce	Received institution-funded training in the previous 12 months
Estonia	86%
Mali health workers	83%
Pakistan Sindh Revenue	74%
Philippines	73%
Madagascar health workers	57%
Uruguay	56%
Argentina, Mendoza	43%
Croatia health workers	43%
Ethiopia	43%
Kosovo	29%
Bosnia and Herzegovina	27%

Source: World Bank, Public Workforce Performance and Prosperity, 2026. Bureaucracy Lab survey periods vary between 2016 and 2025.

These figures should not be interpreted as a league table. The surveys cover different parts of government and were conducted at different times. However, they expose a common structural weakness: many public servants receive little formal development, while others receive training that may not address the problems they face at work.

The World Bank also found that civil servants often need training on specific technical subjects and digital technologies but instead receive training concentrated on laws, regulations and administrative processes.

This is not an argument against understanding government procedures. Those procedures matter. The concern is that administrative knowledge alone does not prepare a public servant to coordinate several ministries, resolve an implementation bottleneck, challenge weak data or turn a political commitment into a funded programme.

Knowledge is not the same as delivery capability

Across government transformation work in Malaysia, the Middle East and Africa, I have seen the same pattern in different forms. A policy can be technically sound but institutionally unowned. A minister may announce a target before the baseline has been agreed. Several agencies may be working on the same issue without sharing one implementation plan. A dashboard may display dozens of indicators but still fail to trigger a decision. These are not problems that can be solved by providing another presentation.

Delivery capability is the ability to operate when the answer is incomplete, authority is dispersed and progress depends on several institutions moving together.

translate a political or strategic priority into measurable outcomes;
separate symptoms from the underlying delivery problem;
establish accountable ownership across institutional boundaries;
connect initiatives to budgets and other required resources;
determine which data can be trusted;
identify dependencies before they become delays;
run a performance discussion that produces decisions;
resolve problems at the lowest practical level; and
escalate only when an issue genuinely requires higher authority.

A participant may understand all these concepts in theory and still be unable to apply them under pressure. This is why public-sector delivery training must be designed as professional practice, not simply knowledge transfer.

What the original PEMANDU experience teaches us

Malaysia’s Performance Management and Delivery Unit, PEMANDU, was established in 2009 to drive implementation of the National Transformation Programme. The operating model concentrated attention on a limited number of national priorities. Those priorities were translated into granular projects with responsible ministries, timelines and key performance indicators.

The World Bank describes PEMANDU’s signature transformation Labs as six-to-nine-week stakeholder workshops that broke each priority into concrete projects. Once implementation began, progress was monitored regularly. Where indicators slipped, implementing agencies were expected to identify the reason, agree corrective action and escalate unresolved problems through the leadership chain.

The World Bank’s independent review found that this combination of prioritisation, granular planning, coordination, monitoring and escalation contributed to PEMANDU’s effectiveness. It also identified important limitations, including data quality, the risk of hitting a target while missing the underlying purpose, and the need for stronger impact evaluation. Read the World Bank assessment of Malaysia’s PEMANDU experience.

Those qualifications are important. A mature PEMANDU training programme teaches the limitations of the model alongside its strengths. Participants need to understand why a methodology worked under particular political, institutional and economic conditions. They also need to recognise when a tool should be adapted, strengthened or abandoned.

The objective is not to produce people who can recite the 8-Step Big Fast Results methodology. It is to develop people who can exercise delivery judgement.

The strongest evidence supports learning through practice

The World Bank’s 2026 report recommends that public-sector training combine classroom and on-the-job learning. It cites rigorous evaluations showing that hybrid programmes using reflective learning-by-doing and team-based learning are more effective than classroom instruction alone.

This aligns with the OECD’s finding that learning and development is a core part of public-sector capability, not merely a product supplied by human-resource departments. Its review also explains how transversal career structures can widen learning opportunities, break institutional silos and create networks that support cross-organisational projects. Read the OECD public-service capability review.

The measurement problem is equally significant.

In February 2026, the Governance Innovation Unit of Bangladesh’s Chief Adviser’s Office published, with technical assistance from UNDP and the Swiss Agency for Development and Cooperation, an evaluation framework for government training institutions. Piloted across five apex institutions, it found comparatively strong infrastructure and training delivery but persistent gaps in outcome measurement, research and evidence of learning impact. It recommended competency-based training, stronger monitoring and evaluation, faculty development and phased institutional reform. Read the government training-institution evaluation framework hosted by UNDP. These findings point in one direction: government training must be designed around work, assessed through application and connected to institutional outcomes.

What PEMANDU-style delivery training looks like

A PEMANDU delivery programme begins with a real implementation problem. Participants arrive with a priority from their own institution: youth employment, investment conversion, hospital waiting times, agricultural productivity, project delays, digital-service adoption or another issue that has resisted ordinary administrative action.

The programme forces that problem through the realities of delivery:

What is the measurable True North?
What is the baseline, target and date?
Which part of the problem can government genuinely influence?
Who owns each intervention?
What must be funded?
Which institution controls the critical decision?
What data can be trusted?
Which dependencies could stop implementation?
What can be resolved operationally?
What genuinely requires political or executive escalation?

Participants leave with a working delivery product, not only course notes. That product contains an implementation plan, named owners, measurable milestones, funding requirements, dependencies, decision rights, performance measures, reporting cadence and escalation pathway.

Simulation is part of the experience. Participants are placed in realistic situations involving incomplete data, institutional resistance, funding constraints, changing instructions and pressure from senior leadership.

This is where trainers must be distinguished from speakers. A strong speaker can explain a transformation story. A competent trainer must be able to diagnose a participant’s problem, challenge the logic, expose missing assumptions and guide the participant towards a usable solution. A delivery coach then supports application after the classroom. An assessor judges whether the person can perform the work independently. These are different professional roles. Treating them as interchangeable weakens training quality.

The PEMANDU Global Delivery Network connects training to professional practice

PEMANDU brings these elements together through the Global Delivery Network. It is not an alumni group, a list of trainers or a catalogue of courses. It is a structured professional pathway that develops government delivery capability from civil-service induction through advanced practice, while connecting practitioners across countries.

The Network operates at three connected levels.

1. Civil Service Induction

The first level introduces new public servants to how government priorities move through Cabinet, ministries, agencies, budgets and frontline institutions. The programme establishes a common language of outcomes, evidence, accountability and implementation. It is adapted to the laws, administrative culture and institutional structure of each country.

A generic international curriculum cannot adequately address local decision rights, political incentives, trust, culture or the relationship between national and subnational government.

2. Delivery Masterclass

The second level is an intensive five-day Delivery Masterclass for delivery-unit personnel, programme managers, senior civil servants and officials responsible for priority initiatives. It is the practical entry point into the Global Delivery Network.

Participants work on live government problems through delivery clinics, simulations, peer challenge and facilitated decision sessions. Their output is assessed against professional standards.

By the end of the programme, our participants have been able to translate a priority into a three-foot implementation plan, construct meaningful KPIs, establish an operating cadence, identify funding and dependencies, and defend the plan under scrutiny.

3. Global Delivery Fellowship

The third level is PEMANDU’s Global Delivery Fellowship for senior practitioners and future government delivery leaders. The fellowship combines taught modules, professional coaching, cross-country peer review and deployment in live delivery environments. Qualification requires evidence of application, not merely the completion of an examination.

A fellow qualifies by demonstrating how they influenced a decision, improved an implementation system, resolved a bottleneck, strengthened institutional ownership or transferred capability to other officials. This creates a pathway from exposure, to application, to professional practice.

The network itself becomes part of the capability

The most useful knowledge in transformation work rarely appears in the final presentation. It sits in the decisions that were made under pressure. It includes rejected options, disputed assumptions, stakeholder resistance, failed interventions and adaptations made when a sound plan encountered political or institutional reality. Most organisations preserve the final report and lose the reasoning behind it.

The PEMANDU Global Delivery Network captures this knowledge while delivery work is happening. Practitioners record:

the original problem and baseline;
the assumptions used;
the options considered and rejected;
the stakeholders and decision rights involved;
the evidence supporting the chosen approach;
the implementation results;
what failed or changed; and
the conditions under which the lesson could be transferred elsewhere.

This creates institutional memory without turning the past into doctrine. A lesson from Malaysia, Rwanda, Oman or Russia is not copied mechanically into another country. It is examined for the political authority, institutional capability, financing, incentives and timing that made it work.

A 2024 World Bank practice note, drawing on a convening of representatives from 15 African delivery units, reaches a similar conclusion. Delivery units can create pockets of effectiveness, improve monitoring and strengthen cross-government coordination, but their institutional form must be adapted to local conditions. Political backing, leadership routines, credible data and ownership by ministries remain essential. Read the World Bank note on African delivery units.

The Inter-American Development Bank examined how delivery units were adapted across 14 Latin American and Caribbean governments and whether they contributed to priority goals between 2007 and 2018. Its research reinforces the point that a delivery unit is an institutional innovation whose performance must be assessed, not an imported template that automatically produces results. Read the IDB technical note.

A data-driven training model measures application

Attendance, completion and participant satisfaction are recorded. They are useful operating measures. They are not sufficient evidence of impact. The Network’s evaluation model covers four levels.

Measurement level	What is tested
Learning	Whether the participant understands the tools, concepts and professional standards
Application	Whether the participant uses those tools on a live workplace problem
Institutional adoption	Whether the participant’s organisation adopts the implementation plan, cadence or decision mechanism
Delivery outcome	Whether decisions move faster, bottlenecks are resolved, data quality improves or the relevant programme outcome changes

Each participant begins with a documented baseline. The programme then conducts 30-day, 90-day and 180-day follow-ups. Not every national outcome can be attributed to training. Economic conditions, political decisions, institutional changes and external events also affect performance. The evaluation therefore distinguishes between direct outputs, plausible contributions and outcomes that cannot be confidently attributed. Credibility depends on that honesty.

AI strengthens evidence, not manufactured confidence

Within the Global Delivery Network, governed AI helps practitioners retrieve comparable cases, identify inconsistencies in implementation plans, generate realistic simulations and compare a participant’s proposed delivery structure against professional standards.

The governed AI layer connects:

programme documents;
decision logs;
delivery dashboards;
approved case studies;
participant assessments;
trainer feedback; and
post-training implementation evidence.

It retrieves relevant lessons with source citations, flags missing ownership or funding, stress-tests assumptions and suggests questions for the practitioner to investigate. It is not allowed to invent evidence, conceal weak data or make unreviewed policy decisions. Human practitioners remain accountable for judgement, political interpretation and final recommendations. The purpose of AI is to help a delivery professional see more, test faster and remember better.

It does not remove professional responsibility.

The real test begins on Monday morning

The purpose of public-sector delivery training is not to make officials sound more sophisticated. It is to make them more capable when implementation becomes difficult.

Can the participant turn an ambiguous priority into measurable outcomes? Can they establish ownership without creating another committee? Can they identify weak data? Can they connect an initiative to funding? Can they run a performance conversation that produces a decision? Can they escalate an issue without surrendering responsibility? Can they adapt a global lesson to local political and institutional reality?

That is the test.

The PEMANDU Global Delivery Network is not built to sell more training days. It develops a recognised profession of government delivery, connects capable practitioners across countries and reduces the long-term dependence of institutions on outside consultants.

Training creates exposure. Practice develops judgement. Professional standards create trust. The network keeps the capability alive.

Author: