How to Build a Dream Team: Five Cases, the Markers, the Checklist
1 DOOR TEAMStrategy & Consulting · 6 September 2026
🌐Статья на английском — перевода пока нет.
I learned management mostly at Mars and LEGO, from leaders who treated it as a craft with rules rather than a personality trait you either have or don't.
And almost every leader I've met since then has underestimated how much deliberate work a good team takes. The bet is on the power of the imperative: say it clearly, say it often, and results will follow.
Unfortunately, it's not that easy. A dream team never assembles itself, and even if you have a team of superstar professionals, it requires even more dedication and work that is mostly unglamorous, invisible from outside, and therefore likely to get skipped. That's the whole subject of this piece.
Below are five situations I have watched live in different companies across the globe. Each one is an act of management negligence, and I am sharing them as a reminder to anyone trying to make a business efficient or take it to scale: the most important component of that success is the team that drives the processes. So this piece covers the markers that tell you which kind of leader you currently are, and the checklist you should use to build your dream team from where you are standing.
If you have direct reports, read the markers section slowly and carefully. If you are just about to have them, save it for later.
The short version (as always)
• The five cases below run from a manager who mixed up his role, did his team's work over a weekend and then blamed them for it, through a CEO who asked a new executive to pass public judgment on a peer, through a culture that asks for loyalty instead of competence, to a forged CFO who had stolen 5 million dollars out of a Fortune 500 company in a few months.
• Toxic culture is ten times more important than compensation in predicting turnover. Gallup's 11th-edition meta-analysis puts top-quartile engaged teams 23% ahead on profitability.
• 82% of people entering management had no formal management training. Gallup estimates managers account for at least 70% of the variance in engagement across business units.
• How 88% of occupational fraudsters pass screening, some of them by pretending to have skills they don't.
Five cases: what breaks a team from the top
For sure, all of this is anonymised and generalised. None describes a single company, and every one will be recognisable to somebody. They run in order of size. The first costs a weekend. The last costs five million dollars.
Case 1. The weekend research
What happened. In one company I watched a manager open the Monday team meeting by announcing that he had spent the weekend on a thorough piece of research. He walked through the findings. Then he told the team, in the same breath, that they should have done this before he did. The work sat squarely inside the team's responsibility. The team had no chance to contribute and was now reacting to a finished artefact, in a meeting. In his model this is how a team operates: read his mind during the week, execute the reading over the weekend.
What it means. The team learned one thing in that hour, and it was not the research. You weren't told about the task, you weren't offered the task, you weren't given the opportunity to participate, but you're definitely the one to blame. So the safe move from now on is to do nothing, wait for the next weekend and react. The manager, having proved he can do the work himself, will keep doing his team's work instead of the work expected of a leader, which is the exact opposite of his job: the team never meets expectations, nobody is happy, and at least the job got done. No surprise that the team in that room turned over completely every twelve months. Churn of 100% a year is a policy, and that meeting is where it gets announced.
What the research says. Google's Project Oxygen, its multi-year study of what separates its best managers from the rest, ranks "empowers the team and does not micromanage" as the second of ten behaviours, above technical expertise (Google re:Work). On the bill: Gallup puts the cost of replacing one person at half to twice their annual salary, so a team that turns over fully each year pays its own payroll a second time in replacement cost alone (Gallup, 2019), and SHRM's 2026 benchmark across 4,600 organisations gives a median 39 calendar days to fill a non-executive role, so each seat sits empty for over a month per cycle (SHRM Recruiting Benchmarking, 2026).
Case 2. Leadership handed over as a title
What happened. In another company the strongest individual contributor on a team was promoted to run the team. Nobody trained him, because the promotion was the reward and training would have implied the reward came incomplete. His working model of management was the only one he had watched from close range: take the plan, pass it down, ask for the number, chase the number. A year later half of the team had left and the other half was burned out. A good performer had suddenly become a nightmare leader: imperative orders instead of vision, goals and support, career conversations that never happened, and the people who had stayed had quietly withdrawn everything that was ever voluntary.
What it means. Leadership had been treated as a title rather than a job with its own skills: running a team, reading motivation, understanding what a group of people does under pressure. Some organisations spend serious money on exactly that, multi-day programmes, coaching, mentoring, turning their best specialists into trained leaders. Others hand over the title and a set of imperatives, then meet the same three problems every time: turnover, low motivation, and the conditional resignation, where a person keeps coming to work as though nothing connects them to the organisation anymore.
What the research says. The Chartered Management Institute, surveying more than 4,500 UK workers and managers with YouGov in 2023, found that 82% of people entering management had received no formal management or leadership training, and that one in three had left a job because of a negative workplace culture (CMI, October 2023). The money, when spent, works: Lacerenza and colleagues pooled 335 samples covering 26,573 people and found leadership training produced large gains in on-the-job transfer and organisational results, with spaced multi-session programmes clearly beating the one-off workshop (Lacerenza et al., *Journal of Applied Psychology*, 2017). And the conditional resignation has a name in the data: Gallup defines not-engaged employees as quietly quitting, psychologically unattached to their work and company, and puts global engagement at 20% (Gallup, State of the Global Workplace 2026).
Case 3. The thumbs-down question
What happened. A newly appointed executive sits through one of his first management meetings. In front of the whole team, the CEO asks him what he would do about one of the country managers, whose team had presented its results a few minutes earlier, and the results were satisfactory. Would you fire him? What would you do in my position? There is no good answer, and the answer was never the point. The point is the demonstration: judgment passed on a peer, in public, on request. And everyone has done the arithmetic in real time. One day the question is about the person beside me, one day it is about me, and the people who will be asked to answer it are sitting here now.
What it means. Everyone in that room learned the same thing in the same second: that chair is a position in an arena, and the rotation is not yours to control. What the company gets afterwards is a group of individually capable people working full time on not being the subject of the next execution. Ownership gets avoided, because ownership is a surface to aim at. Problems get routed towards whoever can carry the blame. Nobody spends their best hours on growing the business, because those hours go on making sure the thumb never turns down in relation to them.
What the research says. Google's Project Aristotle, across 180 teams, found psychological safety, the belief that you will not be punished or embarrassed for speaking up, to be the most important of five factors in team effectiveness, ahead of dependability, structure, meaning and impact (Google re:Work).
Case 4. The culture that asks for loyalty instead of competence
What happened. I have watched this one from inside more than once, and it arrives in the same five moves. The founder becomes unquestionable, and disagreement reads as disloyalty rather than as information. The word "family" surfaces at exactly the moments a family would be irrelevant: pay, hours, boundaries, a request for leave. Curiosity gets handled as a discipline problem. Exhaustion is praised as commitment. And somewhere in there, people stop having a professional identity that exists outside the brand. From the inside none of it feels like control. It feels like belonging, which is precisely the mechanism.
What it means. A company like this has replaced competence with loyalty as the currency it pays in, and the two are not exchangeable. The most capable people are the first to notice, because they are the ones with somewhere else to go. What stays is the people for whom belonging was the point. If you want a diagnostic faster than a culture survey, Jake Breeden's three questions do it in a minute: what happens when someone disagrees; could you explain your work to an outsider without the insider vocabulary; and if you left tomorrow, what exactly would you lose (Psychology Today, 2026). A culture answers the first with a conversation. A cult answers it with a consequence.
What the research says. Teresa Almeida, writing for LSE Business Review, describes how a company's culture can work as a social control mechanism akin to those in cults, where peer influence and social norms recruit and direct members (LSE Business Review, June 2024); the academic version is O'Reilly and Chatman's 1996 paper, blunt in its title, "Culture as social control: Corporations, cults, and commitment". The cost is measurable. Analysing 34 million employee profiles, Donald Sull, Charles Sull and Ben Zweig found toxic culture ten times more important than compensation in predicting turnover (MIT Sloan Management Review, 2022), and in the companion study named the five attributes that drive it: disrespectful, non-inclusive, unethical, cutthroat, abusive (MIT SMR, March 2022).
Case 5. The CFO who was never a CFO
What happened. I met him in his first weeks and something was off. His profile did not read like a person who had spent years inside spreadsheets, pivot tables and reconciliation scripts. He read like a hardened salesman who had moved semi-legal goods to semi-legal buyers. I didn't raise it, because my only evidence was a feeling. Nothing would have happened anyway: "his profile feels wrong" loses to a signed offer every time. He had forged the lot, and he had passed every filter at a Fortune 500 company. For six months he did the ordinary work of the role while gradually rerouting client payments into accounts he controlled. Five million dollars.
What it means. The loud version of the failure is the fraud. The quiet version is the function that let it through: HR, the most underestimated cause of business failures. It is the function where the candidate is always the one at fault, the KPIs are vague and nothing gets a real quality assessment. That leads to wrong hires, no onboarding, no internal path for the people already inside, and searches that run for months while the load sits on sales and marketing. Roughly five out of ten companies lack the competence to write a good job description, posting, offer, onboarding plan or probation review for a technical digital or marketing role. HR is the one function whose failures have no agreed price: the most-quoted cost of a bad hire, 30% of first-year earnings, is attributed everywhere to the US Department of Labor and traces to no primary document I can find. A function that cannot be scored cannot be held to account.
What the research says. The reflex is a better background check, and the data says that is the wrong lesson. The ACFE's Occupational Fraud 2026 report, covering 2,402 cases across 143 countries, found 88% of fraudsters passed pre-employment screening with no red flags, while 84% showed a behavioural red flag once inside and organisations with management review of accounts carried 55% lower median losses (ACFE, Occupational Fraud 2026; ASIS Security Management, May 2026). What would have caught him was verifying the documents themselves: HireRight's 2025 benchmark found three-quarters of employers had uncovered candidate discrepancies in the past year, one in six had experienced identity fraud in hiring, and only three in five run an identity check at all (HireRight, 2025 Global Benchmark Report).
Why a dream team never forms on its own
Each of those cases has a research literature behind it and a price tag attached. Two numbers set the scale. Gallup's State of the Global Workplace 2026 puts worldwide engagement at 20% and the cost of low engagement at roughly US$10 trillion a year, about 9% of global GDP (Gallup, 2026). And in the MIT Sloan analysis of 34 million employee profiles, a toxic culture predicted turnover ten times more strongly than pay (Sull, Sull & Zweig, 2022). The lever is the manager, and it was the manager in every case above.
Direction that moves silently. Locke and Latham's goal-setting work (*American Psychologist*, 2002) found specific agreed goals consistently beat vague or shifting ones. When direction changes without an announcement, effort keeps flowing to the last target given. The work is real; the target moved.
Questions used as tests. Google's Project Aristotle studied 180 teams and ranked psychological safety first among five factors of effectiveness, ahead of dependability, structure and clarity, meaning and impact (Google re:Work). Amy Edmondson's 1999 study in Administrative Science Quarterly found the mechanism: teams that feel safe surface problems earlier and learn faster. An exam question teaches people to hide gaps rather than close them, which is the opposite of what you're paying for.
Doing their job for them. Project Oxygen ranks ten manager behaviours. First is "is a good coach". Second is "empowers the team and does not micromanage" (Google re:Work). Both sit above technical expertise and above results orientation, which surprises most managers who assume competence is what's being scored.
Slow calls on reversible things. Amazon's 2015 shareholder letter separates one-way doors, which are near-irreversible and deserve slow deliberation, from two-way doors, which are changeable and should be decided fast by whoever is closest. McKinsey's 2019 survey of 1,259 executives across 91 countries found 61% believe most of their decision-making time is wasted, and that fast decision-makers are twice as likely to make high-quality decisions as slow ones (McKinsey).
Imperatives instead of context. Netflix's culture memo has held the same line since the 2009 deck: managers "practice context not control - giving their teams the context and clarity needed to make good decisions instead of trying to control everything themselves" (Netflix). Toyota built the same idea into one of the two pillars of The Toyota Way, alongside continuous improvement, and called it respect for people: develop capability deliberately, because it's the engine every other improvement runs on.
Then the money. Modelled on one senior specialist at A$120,000 base, with my own assumptions marked "[estimated]".
Here's the part that makes this hard to catch in time. Disengagement never announces itself. What goes first is discretionary effort: the idea not raised, the problem spotted and not flagged, the thinking that used to happen on a Sunday because someone cared. By the time a resignation lands, that loss has usually been running six to eighteen months.
Three levels, three different jobs
The accountability conversation when something breaks should be hard and direct at every level. The day-to-day gets progressively more hands-off as you move down this table.
Am I a good leader? The markers that answer it
Self-assessment on leadership is close to worthless, because the question "am I a good manager?" gets answered by intention rather than evidence. So here are observable signals instead. No adjectives, nothing that requires you to rate your own empathy.
Good markers
1. Your people bring you problems while they're still small, including ones that make them look bad.
2. Someone on your team disagreed with you in a meeting in the last month, and nothing happened to them.
3. Reversible decisions inside their domain get made without you, and you find out in the update.
4. You can name what each report is specifically better at than you are, without checking notes.
5. Your last three priority changes were announced with the old priority explicitly cancelled.
6. People send you thinking while it's still in progress, before they've defended it internally to anyone.
7. You know your own team's engagement number, not just the company average.
8. When someone on your team gets promoted out, they recommend you to the person replacing them.
Bad markers
1. You hear about problems at roughly the same time the client does.
2. Two of your instructions are live simultaneously and nobody has told you which one they're following.
3. You've done a report's work yourself "because it was faster" more than once this quarter.
4. Your meetings with senior people contain questions you already know the answer to.
5. Your one-on-ones are status updates with a friendlier tone.
6. Nobody has pushed back on you in a meeting for a month, and you've read this as alignment.
7. You can't say when the last decision in your team got a date attached to it.
8. The last resignation surprised you.
Score honestly. Six or more good markers and the system is working. Three or more bad markers and you have a specific, fixable problem, not a character flaw. Every item on that list maps to an item in the checklist below.
My checklist for building a dream team
A checklist is only useful if every line can be marked done or not done by someone other than you. So each item below names the action, the artefact it leaves behind, the pass test, and the case or study it comes from. Sixteen items, in the order I would do them. If you only do the first five, you are already ahead of every manager in the cases above.
In your first week with a team
1. Start with yourself: confidential feedback from the team. Before you assess anyone, get assessed. Use Google's 13 published items, or your own, and remove yourself from the collection entirely: let the team choose the person they trust to gather and merge the answers, or run them through an AI interface that strips names and rewrites phrasing before you see a word. Only where at least three people report to you; below that, anonymity is fiction. Artefact: item-level scores plus the two open answers, verbatim. Pass: within 30 days you tell the team two things you will change and one thing you will not, with the reason. From: Google's semi-annual manager survey; the FAQ above.
2. Get to know the team as people, not functions. One page per person, in your own hand, with four fields: what they are best at in their own words; what matters to them outside work, the interests, the family, the habits, the hour they should never be booked; where they want to be in 24 months; and the thing they have stopped raising. Artefact: the ledger, corrected by each person in their own one-on-one. Pass: for every person you can name, without notes, one thing outside work that matters to them and their 24-month target. From: Gallup Q12 item five, "my supervisor, or someone at work, seems to care about me as a person"; Case 2.
3. Open the decision log and backfill 90 days. Five columns: decision, owner, date, door type (reversible or not), review date. Enter every decision still in force from the last quarter, including the ones nobody wrote down. From now on, any change of direction gets a "supersedes" line naming the instruction it replaces and the date. Artefact: the log, where the whole team can read it. Pass: any team member can name the date of the last three decisions in their area without asking you, and no topic has two live instructions. From: Case 1; Locke and Latham on specific goals; Gallup Q12 item one.
4. Read your baseline before you change anything. Three numbers: regretted attrition on your team over 12 months against the company figure; your team's latest engagement score against the company average; median age of open decisions in the log. Artefact: one line, three numbers, dated. Pass: you can quote all three from memory. From: Gallup's estimate that managers drive at least 70% of engagement variance; Watkins on diagnosing before acting.
5. Ask the stalled-items question and close one. Ask each person: what have you stopped raising because nothing happened last time? Log every answer with an owner and a date. Artefact: the stalled list. Pass: at least one item resolved within 30 days and announced back to the person who raised it. From: Case 3; Google's Project Aristotle on psychological safety.
Every week
6. Praise every person, specifically, every week. No "well done", no "great job". Your job is to find the thing worth praising and describe it precisely: what they did, what it changed, why it was harder than it looked. If you cannot find one for someone this week, that is information about your attention, not their work. Artefact: one written piece of recognition per person per week, somewhere they can keep it. Pass: every piece names a specific action and its effect. From: Gallup Q12 item four, "in the last seven days, I have received recognition or praise for doing good work", which is why the cadence is weekly and not monthly.
7. Brief before you build. Any work you intend to do inside someone's remit gets a written note by Thursday: what you are looking at, why, and which part is theirs. Artefact: the note, sent before the work starts. Pass: zero finished artefacts surfaced to the team unannounced this month. From: Case 1; Project Oxygen behaviour two, does not micromanage.
8. Classify every open decision by door type. Reversible with a bounded downside inside their domain: their call, made this week, logged by them. Irreversible: yours, with a review date. Artefact: door type filled on every log row. Pass: median age of open reversible decisions is seven days or less. From: Bezos, 2015 letter; McKinsey 2019, fast deciders twice as likely to decide well.
9. Run one-on-ones on three fixed questions. Thirty minutes, their agenda, three questions you always ask: what is stuck, what do you want next, what should I stop doing. Status goes in a shared document, never in the meeting. Artefact: running notes with a "stop doing" column. Pass: one "stop doing" answer per month that you actually stopped. From: Google's upward-feedback survey; Gallup Q12 item eleven on progress conversations.
Every quarter
10. Make the career trade explicit. Revisit each person's two-year plan and write it as a two-column deal: what you will do to get them there, and what you need from them and the team in return. Say the second column out loud; people accept a trade far more readily than a favour. Artefact: the two-column plan, in the ledger, dated. Pass: every person can name their next move and what they owe the team for it. From: Gallup Q12 item twelve, opportunities to learn and grow; Google's survey item on career discussion in the last six months; Case 2.
11. Let them try your job. Anyone who wants your position gets a piece of it, for real: "Go to the meeting with my management instead of me. I will help you prepare." A budget review, a vendor negotiation, a board slide. You prep them, you stay out of the room, you debrief after. Artefact: the calendar invite with their name and not yours. Pass: once a quarter for every person who has said they want to grow, and you were not in the room. From: Project Oxygen behaviours one and six, coaching and supporting career development; Lacerenza on transfer beating classroom.
12. Pulse engagement and compare to the company. Twelve questions or a short pulse, read per team, set against the company average and last quarter. Artefact: one chart, two lines. Pass: results and the written "what changes" reach the team inside 30 days. From: Gallup Q12 meta-analysis; the manager-level variance finding.
13. Audit the decision log. Count two things: reversals without a written date, and decisions made above their owner. Artefact: the two counts, quarter on quarter. Pass: zero undated reversals; the escalation share falling. From: Case 1; the cost table above.
14. Check the hiring kit before any role opens. Seven documents exist before the posting goes live: job description, posting, roles and responsibilities, offer template, onboarding plan, probation review, and a written identity and credential verification step. Artefact: the kit folder. Pass: no role opens without all seven; every offer carries a verification line signed off by someone other than the hiring manager. From: Case 5; HireRight 2025; ACFE 2026.
Once, and then hold it
15. Write your never-list and hand it to the team. Three items at most, specific enough to be caught breaking. Mine: never do someone's work over a weekend without telling them first; never ask a question in a meeting whose only purpose is to find out what someone does not know; never ask one person to judge another in front of the room. Artefact: the list, shared. Pass: a team member can quote it back. From: Cases 1, 3 and 4; predictability as the working definition of integrity.
16. Set the altitude per level and stop drifting. Seniors own method, you own outcomes; juniors get direction and cover; managers get context before decisions, not after. Artefact: the three-level table above, with names in it. Pass: you cannot name a method you prescribed to a senior in the last month. From: the altitude table; Netflix, context not control.
Where we land on this
Most owners we talk to describe a strong hire who "isn't delivering". In the five cases above, the person delivering least was the manager, and in four of them nobody had written anything down that a third party could check. So the position we hold, and it is arguable: for a business under 200 people, a decision log, a written hiring kit and a quarterly twelve-question check will do more for output this year than any restructure, any new tool, and most new hires. All three are administrative. All three are nearly free. All three are boring enough to get skipped in favour of things that feel like leadership.
The counter-argument is real. Process does not fix a genuinely wrong hire, and some people who look senior on paper are not. Fine. The log and the kit tell you that faster too, and with evidence, which is exactly what the hard conversation needs.
What changes in 30 days, a quarter, a year
30 days. The decision log exists and gets reviewed in one standing meeting. The hiring kit for your next open role is written before the search opens: job description, roles and responsibilities, offer template, onboarding plan, probation review. Contradictory instructions surface within a week, because the old one is on the page next to the new one.
A quarter. The first engagement read is in, per manager. Reversible decisions are being made by their owners inside a week. You can see whose team score moved and whose did not, and the markers list above has been ticked twice.
A year. Voluntary turnover among senior staff is the number to watch. Gallup's spread puts the gap between top- and bottom-quartile engagement at roughly half your turnover in a low-churn business, and at 0.5 to 2 times salary per departure, that is where the money is.
The uncomfortable part
I'm maybe 70% confident about the ordering of that checklist. The individual items I'd defend to anyone; the sequence is a judgment call and someone with a different team would sort it differently.
What I'd defend without any hedging is the underlying claim, and it's the one most leaders resist: the quality of your team is mostly a description of you. Gallup's 70% figure says it in the language of variance. The five cases say it in the language of a Tuesday afternoon.
That's an uncomfortable read if the team isn't working. It's also the only version of the situation where you have any control at all, which makes it the useful one.
What a good leader actually is
Every serious study of leadership converges on the same four things, and none of them is charisma. Integrity, meaning your actions match your words closely enough that people can predict you. Development, meaning you actively grow the people under you rather than feeling threatened by them. Empathy, meaning you can tell how a message will land before you send it. Accountability, meaning you own your own mistakes out loud, which is the only thing that makes it safe for anyone else to own theirs.
A team doesn't become a dream team because it was told to. It becomes one because somebody did forty boring minutes a week for two years, and never once made an example of anyone.
FAQ
How do I know if I'm a good manager?
Measure it the way large organisations do, with instruments that do not depend on your own opinion. First, a confidential upward-feedback survey: Google runs a 13-item survey on every manager twice a year, answers are anonymised and aggregated, and items include "my manager does not micromanage", "has had a meaningful discussion with me about career development in the past six months" and "I would recommend my manager to other Googlers" (Google re:Work, published via Government Executive, 2017). Second, your team's engagement score read at team level rather than company level, since Gallup estimates managers account for at least 70% of the variance between business units (Gallup, 2015). Third, your team's regretted attrition against the company average. The markers section above is a self-check between measurements, not a substitute for them.
What actually makes a team high-performing?
Two large datasets agree. Google's Project Aristotle, across 180 teams, ranked psychological safety first among five factors, ahead of dependability, structure and clarity, meaning, and impact (Google re:Work). Gallup's Q12 meta-analysis, across 183,806 business units, found top-quartile engaged units 23% more profitable and 18% more productive than bottom-quartile ones, with a true-score correlation of 0.49 between engagement and composite performance (Gallup Q12 Meta-Analysis, 11th edition, 2024). In practice both come down to whether people can say "I don't know" or "this is wrong" without cost.
Why do good employees leave companies that treat them well?
Mostly because of the manager and the culture, and the data is specific. Gallup found 52% of voluntarily exiting employees say their manager or organisation could have done something to prevent them leaving, and 51% say that in the three months before they left nobody spoke with them about their job satisfaction or future (Gallup, 2019). In the MIT Sloan analysis of 34 million employee profiles, a toxic culture predicted turnover ten times more strongly than pay (MIT Sloan Management Review, 2022). Pay is rarely the reason; the absence of a conversation usually is.
What should I do in my first 90 days managing a team?
The standard framework is Michael Watkins's The First 90 Days: diagnose before you act, secure early wins, and build alignment with your own manager on what success looks like. On top of that, four concrete moves from this article: write down what each person is best at and check it with them, start the decision log, hold one development conversation per person, and ask everyone what they have stopped raising because nothing happened last time. Then run the upward-feedback survey at the 90-day mark so the next quarter starts with data rather than impressions.
How often should engagement be measured?
Gallup's Q12 is designed as a full annual survey read at manager level; Google runs its upward-feedback survey every six months. My recommendation for a business under 200 people is the full twelve questions once a year plus a short pulse each quarter, both reported per manager and both followed inside 30 days by a written "here is what changes". The follow-through matters more than the interval: a survey with no visible response trains people to stop answering honestly.
If you're reading the bad markers and recognising three of them, that's a process problem with a known fix, and it's a good week to start. If you'd rather have someone map it with you, write to me. We run the same structured audit on teams and decision flow that we run on websites, and the first conversation costs nothing.
Comments
No comments yet. Be the first.