How people
grow at work.

Better programs begin with a better understanding of people.

A workshop can be well attended, beautifully presented, and quickly forgotten. The harder question is what someone can do differently when they return to work, and whether the organization makes that difference possible.

Learning, leadership, performance, and culture are often managed as separate programs. People experience them together. A manager learns to invite feedback in a classroom, then returns to a meeting where disagreement is punished. An employee learns a better process, but has neither the authority nor the time to use it.

This study brings adult learning principles into conversation with research on how teams work. First, the learnings. Then, frameworks for designing people programs. Finally, a few personal musings on what those frameworks ask of the people doing the designing.

The central proposition (editorial synthesis): Design for a meaningful attempt, a useful response, and another chance to practice. Design the surrounding work to support all three.

This is an independent research synthesis and proposed design approach, not a delivered client engagement, a systematic review, or evidence of measured program outcomes. The frameworks and illustrations are design aids, not a newly validated theory.

01 / The learnings

Adults arrive with a life, not a blank page.

Malcolm Knowles’ andragogy shifts attention from delivering material to understanding the learner. His 1973 account foregrounds self-direction, accumulated experience, readiness tied to life roles, and a problem-centered orientation. It also explicitly resists a clean division between how children and adults learn. These are assumptions to examine, not biological rules that switch on at adulthood. [16]

Later accounts commonly organize andragogy into six assumptions, adding the need to know why and motivation to learn. A CDC-hosted practical guide offers a related six-principle formulation emphasizing motivation and self-direction, experience, goals, relevance, practical application, and respect. The lists overlap; they should not be presented as one unchanged original checklist. [17], [22]

For program design, the useful questions are straightforward:

• Purpose: What real problem will this help someone solve? Explain the benefit before the syllabus.
• Agency: Where can the learner choose a case, a goal, or a route through the material? Offer guidance without confusing autonomy with abandonment.
• Experience: What do people already know? Use their examples, while leaving room to question habits that no longer serve them.
• Readiness: Why does this matter now? Connect the learning to an actual responsibility or transition.
• Application: What will someone practice doing? Organize around decisions and situations, not only topics.
• Motivation and respect: What matters to the learner? Make development useful without assuming everyone has the same ambition, confidence, or access to opportunity.

These questions are an editorial translation of the sources, not six experimentally established effects. Adults vary. Novices may need explicit instruction and worked examples; experience can support learning or carry misconceptions. Choice should sit inside clear expectations, especially where safety, conduct, or compliance requirements are non-negotiable.

Learning needs somewhere safe enough to be unfinished.

Asking a question, admitting confusion, or trying a new approach exposes a gap. If revealing that gap is costly, the program may receive polished participation instead of useful information.

Amy Edmondson’s 1999 study linked team psychological safety with learning behaviors such as seeking feedback and discussing errors. Google’s Project Aristotle and a later meta-analysis extend the conversation beyond that original setting. Much of this evidence is associational; it does not establish that one facilitation technique will cause better performance everywhere. [1], [2], [3]

Psychological safety is an important enabling condition for interpersonal learning, not a guarantee of learning, a demand for comfort, or an exemption from accountability. In design terms: give people a low-stakes place to rehearse, respond constructively to mistakes, and make the standard visible. Do not use someone’s vulnerable practice attempt as undisclosed performance-evaluation evidence.

Experience becomes learning when the next attempt changes.

Kolb’s experiential learning cycle offers a useful structure: concrete experience, reflective observation, abstract conceptualization, and active experimentation. Try something, examine what happened, form an explanation, and test a different approach. [18]

A leadership session might therefore begin with a difficult conversation, not a lecture about communication. Participants rehearse, notice the effect of their choices, compare those observations with a model, and try again. This is an illustrative adaptation, not evidence that this particular sequence outperforms every alternative.

The cycle is a planning lens, not a reason to label employees by a fixed learning style. Reflection alone can also rehearse a poor explanation; accurate examples, knowledgeable feedback, and checks for understanding still matter.

Remembering requires more than recognizing.

Dunlosky and colleagues’ review rated practice testing and distributed practice highly across a range of learning conditions. Retrieving an idea from memory and returning to it across separated sessions can support durable learning better than relying only on rereading or a single concentrated exposure. [19], [20]

For a workplace program, that suggests short, low-stakes recall prompts, fresh scenarios, corrective feedback, and later opportunities to apply the idea. The evidence is strongest for learning and retention tasks; it does not automatically establish an effect on leadership quality, employee engagement, or organizational performance. There is no universal spacing schedule for every skill.

Teams shape what individual learning can become.

The earlier team research adds four useful constraints to individual learning design:

• Feedback needs an owner. Pixar’s Braintrust separates candid advice from authority over the final decision. A peer review can diagnose a problem without becoming another approval gate. The evidence here is a practitioner account, not a controlled evaluation. [4], [5]
• Autonomy needs context. McChrystal’s account couples shared understanding with empowered execution; Willink and Babin emphasize ownership and coordination. These are selected leadership narratives, not proof that military practices transfer unchanged to civilian organizations. [6], [7], [8]
• Participation is not an employee score. Woolley and colleagues associated group performance with social sensitivity and conversational participation. Later affirmative and critical syntheses disagree about the strength and interpretation of collective intelligence. Pentland’s field account offers another perspective on communication patterns, not a universal productivity formula. Surface relevant knowledge; do not rank people by airtime or infer ability from gender. [9], [10], [11], [12]
• An adopted tool is not an implemented practice. Haynes’ surgical checklist study reported improved outcomes after implementation; Urbach’s Ontario evaluation did not find a statistically significant reduction in mortality or complications. The settings, endpoints, baseline risks, and implementation differed. Both were observational before-and-after studies, and the checklist includes clinical checks as well as communication. The contrast is a warning against equating completion with effectiveness, not a claim that culture alone explains the difference. [13], [14], [15]

What carries across (editorial synthesis): Capability is partly individual and partly environmental. A person needs something worth learning, a credible way to practice, useful feedback, and room to use the learning afterward.

02 / The frameworks

People programs as products. People are not the product.

The useful borrowing from product practice is the loop: understand a need, make a focused intervention, observe use, and improve it. The boundary matters just as much. Employees are not conversion targets. Confidentiality, dignity, equitable access, and the limits of measurement belong in the design from the start.

The following loop is my editorial synthesis of the research, not a tested package.

1. Discover the need, not just the requested course.

Start with a concrete work situation. Combine employee feedback, interviews, work samples, and appropriately aggregated engagement or performance data. Ask whose experience is missing, including people with less schedule flexibility or less power to speak openly.

Distinguish a skill gap from unclear roles, poor tools, conflicting incentives, excessive workload, or lack of authority. Training is not the default fix for a system that makes good work difficult.

Design output: a short problem statement naming the audience, desired behavior, current barriers, and intended outcome. State what evidence would change the diagnosis. Explain survey confidentiality honestly; do not promise anonymity where a small group or a free-text response could identify someone. Collect only what is needed, restrict access, and agree retention rules.

2. Design the attempt, not only the content.

Use the adult learning questions to choose a relevant problem and an observable learning objective. Build an experience–reflection–explanation–retry sequence. Give beginners enough instruction to attempt the task and experienced participants enough challenge to reconsider their defaults. [16], [17], [18]

Connect formal instruction, peer learning, and supported work practice. The familiar 70–20–10 formulation (experience, learning with others, and formal learning) is a heuristic with weak empirical support for its precise percentages, not a measured law or a budget allocation rule. Use it to notice missing connections, not to prescribe a ratio or justify withdrawing instruction. [21]

Design output: a practice task, clear quality criteria, an accessible job aid, a feedback method, and opportunities to retrieve and reapply the learning later. The amount and timing should fit the task, not a borrowed percentage.

3. Pilot the whole experience, including the return to work.

Try a manageable version with a defined group. Explain what is being tested and how practice data will be used. Provide accessible materials and alternatives for participation. Prepare facilitators to handle disagreement without humiliation.

Ask managers to protect practice time and create an actual opportunity to use the skill. Identify the program owner, the sponsor who can remove barriers, and the person responsible for follow-through. A rollout needs those commitments, not simply a launch announcement.

Design output: a pilot plan with decision criteria, support responsibilities, and a review point. Look for confusion, exclusion, workload costs, and unintended consequences, not only enthusiasm. A willing pilot group may not represent the rest of the organization.

4. Measure the difference without overstating the cause.

Kirkpatrick’s four levels distinguish reaction, learning, behavior, and results. Plan backward from the intended result, then identify the behavior and learning needed to support it. [23]

• Reaction: Did participants find the experience relevant and usable? Ask about barriers, not only satisfaction.
• Learning: Can they demonstrate the intended skill? Use an appropriate scenario or task with clear criteria; confidence alone is not competence.
• Behavior: Does the practice show up at work? Examine consented observations or suitable work samples, alongside whether people had a chance to apply it.
• Results: Did the organizational outcome move? Pair the desired outcome with quality, workload, and fairness checks so improvement is not achieved by shifting costs elsewhere.

The levels organize questions; they are not a causal staircase. A favorable reaction does not prove learning, and a business metric moving after a program does not prove the program caused it. Define baselines and measures before rollout. Where feasible, use a credible comparison or staggered introduction and record other changes that could explain the result. Keep conclusions proportionate to the design.

5. Iterate the program and the conditions around it.

If learning improves but behavior does not, investigate opportunity, incentives, tools, and manager support before adding another module. If behavior changes without the intended result, revisit the original diagnosis.

Bring a short account back to participants: what was heard, what changed, what did not, and why. Name an owner for the next revision. Performance management can reinforce the loop through clear goals, useful coaching, and recurring development conversations, while keeping developmental practice distinct from formal ratings.

Design output: a decision to adapt, expand, pause, or stop, with reasons. A program that should end is not a failed reporting exercise.

A first design: practice useful feedback.

Illustrative program concept. Not a real engagement or a claim of results.

Begin with a bounded behavior: a manager can describe an observable issue, explain its effect, invite the employee’s perspective, and agree a next action. Discover whether the obstacle is the skill itself or the conditions around feedback before proceeding.

Invite participants to bring an anonymized situation or choose a supplied scenario. Explain why the skill matters. Model a conversation, rehearse it, use a shared rubric for peer feedback, and retry. Return to a new scenario later; ask participants to recall the approach before consulting the job aid. Pair this with a supported opportunity to use the skill at work. This combines andragogy, Kolb, and spacing/retrieval as a design hypothesis, not a validated curriculum.

Evaluate relevance during the pilot; compare demonstrations against the same criteria before and after; later examine appropriate evidence of workplace use. Agree the organizational outcome with stakeholders rather than inventing a target. Do not ask employees to disclose sensitive conversations or treat candid practice as material for a rating.

Before the first invitation goes out

• Can we name the work problem and the people affected, without making a course the answer in advance?
• Have we checked whether the barrier is knowledge, skill, opportunity, or the system around the work?
• Is the objective observable, relevant, and understandable to the learner?
• Have we made room for prior experience, supported choice, accessible participation, and explicit instruction where needed?
• Is there a realistic attempt, useful feedback, a retry, and later practice?
• Have managers committed time, opportunity, and support after the session?
• Are the baseline, success criteria, data protections, and evaluation limits agreed?
• Does someone own the decision to improve, expand, or stop the program?

03 / Musings

Personal reflections. These are opinions, not quotations, personal anecdotes, or empirical findings.

Relevance is a form of respect.

I think a learning program should be able to explain why it deserves a place in someone’s day. Not with a grand promise about transformation, but with a recognizable problem and a useful next step. Respect starts before the first slide.

The environment is part of the curriculum.

I’m less interested in teaching people to speak up if the next meeting teaches them to stay quiet. The workshop and the workplace are both giving lessons. Good program design asks whether they are teaching the same thing.

Frameworks should help us notice, not stop us thinking.

I like a framework that makes a missing piece visible. I’m wary of one that makes a complicated situation look settled. A neat percentage or a clean diagram can be a useful beginning. It should not become the reason we stop asking questions.

The second attempt deserves more attention.

The first attempt tells us what someone can do today. The second tells us whether the experience helped. I’d rather design a smaller program with a meaningful retry than a larger one that leaves no room to use what was taught.

Taste belongs in systems, too.

To me, taste is not just how a program looks. It is knowing what to leave out, when to offer structure, and where to give someone room. A thoughtful program should feel clear and useful, not like another process people have to work around.

Notes & sources

Selective research synthesis. Sources 1–15 carry forward the earlier team study; they were reviewed through that study and its saved underlying materials, not all reopened online for this edition. New learning sources were checked online for this edition, with access limits noted beside each source. No invented figures, effect sizes, quotations, testimonials, or client outcomes. All diagrams are editorial adaptations and show no measured quantities. Framework-provider descriptions explain their models; they do not independently establish effectiveness.

1. Understand team effectiveness

   Team-research source carried forward; see evidence limits in the text.

2. Psychological Safety and Learning Behavior in Work Teams

   Team-research source carried forward; see evidence limits in the text.

3. Psychological Safety: A Meta-Analytic Review and Extension

   Team-research source carried forward; see evidence limits in the text.

4. Inside Pixar's Braintrust

   Team-research source carried forward; see evidence limits in the text.

5. Questions for Ed Catmull

   Team-research source carried forward; see evidence limits in the text.

6. Extreme Ownership

   Team-research source carried forward; see evidence limits in the text.

7. Adapt to Win in the 21st Century

   Team-research source carried forward; see evidence limits in the text.

8. Team of Teams: Executive Summary

   Team-research source carried forward; see evidence limits in the text.

9. Evidence for a Collective Intelligence Factor

   Team-research source carried forward; see evidence limits in the text.

10. The New Science of Building Great Teams

    Team-research source carried forward; see evidence limits in the text.

11. g versus c: two meta-analyses

    Team-research source carried forward; see evidence limits in the text.

12. Quantifying collective intelligence in human groups

    Team-research source carried forward; see evidence limits in the text.

13. A Surgical Safety Checklist to Reduce Morbidity and Mortality

    Team-research source carried forward; see evidence limits in the text.

14. Introduction of Surgical Safety Checklists in Ontario

    Team-research source carried forward; see evidence limits in the text.

15. Surgical Safety Checklist: Implementation Manual

    Team-research source carried forward; see evidence limits in the text.

16. Malcolm S. Knowles · The Adult Learner: A Neglected Species (1973)

    Original text, especially pp. 43–49; four early assumptions and the warning against a strict adult/child divide. PDF downloaded and text checked.

17. Adult Learning Principles · CDC-hosted practical guidance

    A related six-principle formulation, not a reproduction of the original 1973 list. PDF downloaded and read.

18. Institute for Experiential Learning · What Is Experiential Learning?

    Framework-provider explanation of Kolb’s four stages; page opened. The study does not endorse learning-style matching.

19. Dunlosky et al. · Improving Students’ Learning With Effective Learning Techniques (2013)

    Review abstract verified in web search; direct fetch rate-limited. Accessible author explanation read in source 20.

20. John Dunlosky · Strengthening the Student Toolbox (2013)

    Author’s explanation of the review, including retrieval, distributed practice, corrective feedback, and limits; full page opened.

21. University of Twente · Rethinking the 70–20–10 Rule (2026)

    University teaching guidance reviewing the weak empirical basis of the ratio; page opened. A secondary critique, not a trial.

22. Andragogy in Practice: Applying a Theoretical Framework to Team Science Training in Biomedical Research

    Peer-reviewed application explaining six later assumptions; abstract and table checked in search. Direct fetch blocked; public curl reached a browser-check page. No effectiveness or numerical claims taken from this paper.

23. Kirkpatrick Partners · The Kirkpatrick Model

    Official commercial framework description; page opened. Used for the four-level taxonomy, not as independent proof of effectiveness.