The work of helping people grow

A manager has an unusual degree of influence over another person’s life. They help decide which opportunities that person receives, whose account of their work is believed, and whether a difficult period becomes something they recover from or a reputation they struggle to escape. Much of this influence is exercised through ordinary decisions about assignments, feedback, and where to spend an hour of attention.

I think good management requires us to take that influence seriously while remaining responsible for the work a team has promised to do. We want people to succeed now and become capable of more over time. Those ambitions can come into conflict when a deadline is close, someone is struggling, or helping one person consumes time another person needs. Caring about the team does not spare us those choices. It makes the quality of our choices more important.

A high standard should come with a clear account of how people can meet it, and a manager willing to be examined on the support, opportunities, and decisions they provide. Without that second part, we can become demanding evaluators of development we have done very little to enable.

Understand what you are responding to

Suppose a product manager has missed two agreed milestones. A colleague describes them as disengaged. Before accepting that description, I would want to understand what happened to the work and hear the person’s account. They may lack a skill, be carrying conflicting assignments, have misunderstood their authority, or be dealing with something outside work. They may also have failed to follow through on a reasonable commitment. The missed dates establish a problem; they leave its explanation open.

The explanation matters because the responses are different. Teaching someone how to plan will accomplish little if three leaders keep replacing their priorities. Removing responsibilities may relieve an immediate problem while denying them the practice they need. Repeating that we believe in them offers little direction when they genuinely do not know how to proceed.

An attentive conversation begins with something concrete and leaves room for correction. The Center for Creative Leadership’s feedback approach separates the situation, observed behavior, and impact, then asks about the person’s intention.¹ A conversation might begin, “The research sessions were postponed twice, and design is waiting for the findings. Where did you get stuck, and what have I missed?” The answer should be able to change my understanding, including my assessment of whether I gave them a workable assignment.

Personal circumstances require particular care. A person should be able to ask for support without being pressed to disclose the details of their private life. Where a temporary change in workload or time away is appropriate, we need to agree what changes and arrange coverage. Colleagues need clarity about responsibilities; they do not need access to the personal explanation behind them.

Listening also continues after the first conversation. Someone can be upset by feedback and still use it thoughtfully once they have had time to consider it. I would look at their response over time and the reasoning behind any disagreement. Requiring immediate agreement would make it harder for them to correct an inaccurate account of their own work.

Make the standard something a person can learn

Once we understand the gap, the next question is whether our feedback gives the person a way to improve. “You need more executive presence” could refer to an unclear recommendation, an unsupported assertion, or simply a style of speaking the manager does not prefer. Those are very different concerns. We owe the person an explanation of the specific behavior that matters and why.

Imagine a team member who accepts an additional commitment during a planning meeting without examining what it displaces. The coaching can focus on that moment. Before the next meeting, work through the existing commitments, practice explaining the conflict, and agree which decisions they can make themselves. Afterward, examine how they handled the trade-off and what still proved difficult. The assignment remains theirs, but they have received something more useful than an instruction to show greater ownership.

This takes time, and it can be tempting to take over. A manager can rewrite a document, lead the difficult meeting, and deliver a stronger immediate result while leaving the employee dependent on the same intervention next time. Sometimes the consequences require us to step in. When they do, we should explain why, review the decision afterward, and give the person another appropriately bounded opportunity to exercise the skill. Repeated rescue should prompt us to examine whether our coaching is producing independence.

The opposite mistake is withholding useful direction in the name of empowerment. Someone doing unfamiliar work may need an example, a demonstration, or an early review before they can exercise meaningful autonomy. The amount of support should change with their experience and the consequences of error. A strong practitioner may chiefly need a clear objective and freedom over the approach; someone developing the capability may benefit from closer checkpoints that become less frequent as their work improves.

We should be equally specific about good work. CCL’s feedback method applies to useful behavior as well as problems.¹ When someone brings two disagreeing colleagues to a workable decision, tell them which part of their approach helped. Recognition gives them a clearer understanding of what to repeat, and making the contribution visible helps ensure that the person who enabled the result receives credit for it.

Raising the standard then becomes a change we can explain in advance. Perhaps the next assignment requires handling a wider set of consequences or making decisions with less intervention. It should come with an appropriate opportunity to practice. Discovering at review time that yesterday’s agreed success has become today’s inadequacy is a poor substitute for development.

Take ambition seriously enough to examine it

The same obligation to make feedback useful applies when someone brings us an ambitious idea. I would never use “pie in the sky” as the explanation for rejecting it. The phrase gives the person a verdict without identifying whether the problem is technical feasibility, weak evidence of benefit, cost, timing, or a change the organization has not yet learned how to make. Each would require a different response; some might expose a limit in our own understanding.

Suppose a team member proposes using AI to coordinate a process that currently passes through several departments. I would ask them to explain the outcome, which decisions still require specialist involvement, and the assumption most likely to prevent the approach from working. Where there is a credible opportunity, help them reach the right experts and define a test that can answer the important uncertainty. A bounded test should preserve the idea’s potential advantage rather than shrink it into a familiar feature that tests something else.

A proposal may merit rejection because its plausible benefit cannot justify the cost, or because another investment matters more. Explain that judgment and give the person a chance to challenge its basis. If we ask for more evidence, we should identify what could change the decision and make time for the investigation, rather than leave them pursuing approval we have already ruled out.

The developmental consequence matters. People learn what kinds of thinking we are prepared to hear. If ambitious work is answered with a dismissive label, I worry that they will begin editing it out before we see it. We can ask for greater rigor while helping people develop the ability to pursue larger ideas. That requires us to show our own reasoning, acknowledge where we lack expertise, and be willing to support an answer that differs from the one we expected.

Give growth a place in the work

Development also depends on which work people receive. A reliable person can become the default owner of every difficult recovery because the manager knows they will get it done. That may help the team through the current quarter while preventing that person from pursuing the work they need for the next stage of their career.

I would discuss four things together: what the person does well, which parts of the work give or drain energy, where their contribution matters, and what they want to learn. These are questions to explore with them. A strong analyst might want broader responsibility; another might want to deepen their expertise. We should not assume that progress means the same title, scope, or managerial path for everyone.

The useful detail often lies inside a task. Someone may enjoy speaking with customers but find arranging interviews and reconciling scattered notes exhausting. Another may be comfortable with difficult conversations but drained by meetings in which settled decisions are repeatedly reopened. Calling either person “bad at research” or “weak with stakeholders” would conceal a possible change to the work itself.

Enjoyment cannot determine every assignment. Gagné and Deci’s account of work motivation distinguishes doing something because it is intrinsically interesting from other forms of motivation that can still be relatively autonomous.² A person can willingly undertake demanding, unexciting work because they understand and value its purpose. As managers, we should explain that purpose, consider how the burden is distributed, and avoid treating someone’s willingness to help as an unlimited resource.

A development opportunity needs room in the schedule. Giving someone a stretch assignment while leaving every existing obligation intact asks them to finance their growth through additional hours. I would agree what comes off their plate, who can help, and what responsibility they will actually have. The colleague providing support needs time for that contribution too.

Career conversations require the same honesty. If someone is ready for greater responsibility, make the case and help them get access to the work. If the organization has no suitable opening, say so rather than invent another personal deficiency. Readiness and the availability of a role are different questions. Our obligation is to advocate accurately and follow through on the commitments we can make, including when a person’s best next opportunity sits outside our team.

Be accountable to the whole team

Supporting an individual becomes unfair when its costs are left invisible to everyone else. If colleagues repeatedly repair missed work, cover undeclared absences from shared responsibilities, or tolerate conduct that makes collaboration difficult, the manager needs to address the arrangement. Quietly relying on the most dependable people to absorb it is itself a management decision.

The standard should include how results are achieved. A person who delivers valuable work while repeatedly belittling colleagues creates a different problem from someone who raises an uncomfortable but well-supported objection. We need specific examples to distinguish the two. Popularity, similarity to the manager, and the absence of complaints are inadequate substitutes for understanding what happened.

Amy Edmondson’s study of 51 work teams found that psychological safety was associated with learning behavior. This is the shared belief that the team is safe for interpersonal risks such as admitting a mistake or asking for help.³ I take the practical implication to include the manager’s response when someone admits an error or questions a decision. We need the information early enough to act, and should examine the reasoning and circumstances rather than treating disclosure itself as evidence against the person. A team still needs dependable work and respectful conduct; it also needs a way to reveal when either is missing.

When a serious performance gap persists, the path forward must remain honest. State the change required, the support available, and when progress will be reviewed. Record the manager’s commitments alongside the employee’s. Before concluding that development has failed, check whether the promised coaching, access, and protected time actually arrived. The person should know the seriousness of the situation while there is still a genuine opportunity to respond.

There will be cases where the essential requirements remain unmet despite meaningful support. A different role may be appropriate where there is a real business need and a credible fit; in other cases, a difficult employment decision may follow. We should handle that through a fair process and with dignity. Presenting a decision already made as an open coaching exercise denies the person an honest understanding of their position.

Our own performance belongs in the same examination. If I repeatedly cancel the conversations meant to support someone’s development, or add work without resolving priorities, that should affect how my leadership is assessed. When I have misunderstood an event, I should correct the account with the people who heard it. An apology confined to the employee does not repair a mistaken impression I helped create elsewhere.

Use AI to improve the attention we give

AI could make some of this work easier, especially the preparation and follow-through that busy managers neglect. I would start by using it to examine my own reasoning. Given a carefully minimized, authorized account of an interaction, it could help separate observations from assumptions, identify questions left unanswered, or flag feedback that describes a personality rather than a behavior. The useful output is something to investigate with the person.

There is a particular risk when the manager supplies the entire account. Anthropic’s research on sycophancy found that the assistants studied could favor responses matching a user’s beliefs over more truthful ones.⁴ Asking a model to explain why an employee is difficult can therefore produce a polished extension of the assumption we should be examining. I would instead ask what in my account is unsupported, what reasonable alternative explanation I have omitted, and what I should ask before drawing a conclusion. The employee’s own account remains essential.

Another opportunity is to look for patterns across the work people do, rather than rely on whichever experience they remember during a one-to-one conversation. I would experiment with a couple of weeks of optional, private reflection in which people note the task, what it demanded, whether it gave or drained energy, and where they believed it created value. The person could use an approved AI tool to compare those entries and prepare a summary they choose to share. Perceived impact would remain distinct from evidence that an outcome improved.

Suppose those entries suggest that writing is consistently draining. Comparing the examples might show that the person enjoys developing an argument but struggles when it reaches a review whose expectations keep changing. Another person may welcome a demanding review and struggle to get started without a clear question. The useful output would identify the possible pattern, the examples supporting it, and occasions that do not fit. The person needs to be able to correct it, and sparse or contradictory entries may support no conclusion.

The next step would be a joint test of what might help. The first person might need clearer review criteria; the second might benefit from coaching on breaking an open question into tractable ones. If handling disagreement is central to the role, the answer may be deliberate practice with support. We would agree on a small change and a review date, then look at the person’s experience alongside the quality and follow-through of the work. A recurring pattern supplies another signal; it does not establish the cause.

With permission to combine the summaries people have chosen to share, AI could also help a manager examine the distribution of work across the team. It might flag that the same person repeatedly absorbs interruptions while others receive assignments that develop new skills, or suggest pairings that combine an experienced colleague’s strengths with someone else’s learning goal. I would compare those suggestions against business needs, actual work, and what each person wants to develop. Essential but draining work still needs a fair owner; enjoyment should not become a rule that confines people to yesterday’s strengths.

I would judge the experiment by whether it improves a concrete coaching or allocation decision enough to justify the effort, including the time people spend recording and correcting it. A better-sounding description of the team would be insufficient if our assignments, support, and follow-through stayed unchanged.

The data arrangement needs to match that promise. The raw reflections should remain outside the manager’s access, with an ordinary conversation offering the same opportunity to ask for support. Decide what may enter the tool, who can access it, how long it is retained, and whether it can be reused. Combining summaries from a small team does not guarantee anonymity. Exploratory reflections should not quietly become inputs to ratings or promotion decisions, and a generated profile should not become an enduring description of someone’s capability.

AI can also provide a place to practice. Someone preparing for a difficult conversation could rehearse several plausible responses, then examine whether their message is clear and leaves room for disagreement. A product manager could practice explaining a trade-off and investigate a challenge they struggled to answer. These are rehearsals to learn from; the simulated colleague does not tell us what the real person thinks.

We should still check whether the practice develops understanding. In a randomized study of developers learning an unfamiliar programming library, the AI-assisted group scored lower on an immediate comprehension assessment, with no statistically significant improvement in completion speed.⁵ The study concerns a bounded coding task, but it gives managers a reason to assess what someone can explain and adapt after using a tool, alongside the quality of the assisted output. A completed artifact alone leaves that question unanswered.

The opportunity I find most useful is to notice more about the work while giving more attention to the person doing it. An assistant might surface a recurring obstacle, help track an introduction I promised, or prepare questions for a development conversation. That would be worthwhile if I then investigate the obstacle, make the introduction, and work with the person to change what needs to change.

A good manager should make it easier for people to do demanding work without having to hide where they need help. Over time, I would hope to see them take on decisions that once required our intervention, help colleagues develop in turn, and recognize possibilities we would not have found ourselves. We should ask whether our assignments, feedback, and use of time are making that future more likely, then change the parts that are getting in its way.

---

Notes

Footnotes

1. Center for Creative Leadership, “Use Situation–Behavior–Impact (SBI) to Understand Intent” (2025), and “SBI Feedback Model & Talent Development Conversations”. The organization’s method addresses specific situations, observed behavior, impact, and inquiry into intent; it can be used for both positive and corrective feedback. The examples and proposed management practices in this post are illustrative. Back to reference 1 Back to reference 1-2

2. Marylène Gagné and Edward L. Deci, “Self-determination theory and work motivation”, Journal of Organizational Behavior 26(4), 331–362 (2005). This is a theoretical account of work motivation distinguishing forms of motivation by their degree of autonomy; it does not imply that enjoyment alone determines good work allocation. Back to reference 2

3. Amy Edmondson, “Psychological Safety and Learning Behavior in Work Teams”, Administrative Science Quarterly 44(2), 350–383 (1999). The study examined 51 teams in a manufacturing company. The association with learning behavior informs the discussion here; the article’s management proposals are applications of that concern, rather than interventions tested by that study. Back to reference 3

4. Mrinank Sharma and colleagues, “Towards Understanding Sycophancy in Language Models” (2023). The researchers studied five AI assistants across four free-form generation tasks. The findings establish a failure mode to test for, rather than a measured error rate for current workplace-management tools. Back to reference 4

5. Judy Hanwen Shen and Alex Tamkin, “How AI Impacts Skill Formation” (2026); researchers’ account. The main study involved 52 developers learning an unfamiliar Python library and assessed comprehension shortly afterward. The observed relationship between different interaction patterns and learning was exploratory rather than causal; the study does not establish long-term effects or outcomes for people-management coaching. Back to reference 5