Psychological Safety Is Not Soft Management
Psychological Safety Is Not Soft Management
On this page
On large capital projects, the information that decides outcomes usually exists long before it reaches a decision. An estimator suspects that a productivity assumption is optimistic. A site engineer notices that a supplier's test results are drifting. A package lead sees that two interfaces were drawn to different revisions. Each of them runs a quick private calculation about what will happen if they say so — and on many projects, the answer is that saying so will cost more than staying quiet. Owners who want that information to travel need a particular kind of team climate. The common name for it is psychological safety, and the common mistake is to treat it as soft management. It is the opposite: a climate in which candour is cheap and standards stay high.
1. The Owner Who Thought Safety Meant Softness
Picture a sponsor on a major programme being told that the project team needs more psychological safety. The phrase lands badly. It sounds like fewer hard conversations, gentler design reviews, missed dates excused in the name of wellbeing, and a leadership style that has stopped asking for results. On a job with a fixed sanction value, a regulator watching and a market window closing, that sounds like a luxury the project cannot afford.
The sponsor's instinct to protect standards is right. The diagnosis is wrong. The real problem on most large capital projects is not that people are under too much pressure to perform. It is that people who already know something important decide that raising it is not worth the personal price. The estimator who doubts the productivity figure has watched the last person who challenged the baseline get labelled as negative. The engineer with the drifting test data knows the supplier relationship is commercially sensitive and that bad news tends to be treated as a sign of weak management of the package. The package lead with the mismatched interfaces suspects that admitting the mismatch means admitting a mistake in their own coordination.
None of these people lack competence or commitment. They are doing exactly what people in every organisation do: asking themselves, often without noticing, whether speaking up will make them look ignorant, incompetent, negative or disruptive. When the answer is yes, the information stays where it is. It surfaces later, through test failures, claims, rework and schedule slippage, by which time the cheap options have gone.
The space-programme record offers a well-known illustration. Where a commercial climate became punitive, contractors held back troubling test data rather than share it with a customer whose reaction they feared. The data did not disappear. It simply arrived late, when it was most expensive to act on. Large capital projects reproduce the same pattern whenever reporting bad news is treated as an admission of failure.
The point of this essay is therefore narrow and practical. Psychological safety is about the price of candour, not the price of failure. A project can make candour cheap while keeping the price of poor performance high. Treating those as the same dial is what makes safety look soft.
2. Where This Argument Sits
Psychological safety sits underneath a cluster of related ideas, and it is easy to blur them. This essay keeps a deliberate boundary so the core argument stays visible.
It is not an argument about how to stage productive disagreement or how to draw buried conflict to the surface in a meeting. Safety is the climate that makes debate affordable; the techniques for running debate are a separate matter.
It is not an argument about silence being misread as consent in decision meetings. Quiet rooms are one symptom of low safety, and they appear here only as a symptom.
It is not an argument about rank, airtime and the way hierarchy compresses discussion. Power appears here only as one lever a leader must manage.
It is not an argument about personal trust between team members, which builds over time and rests on willingness to be vulnerable with colleagues. The distinction between trust and safety is drawn once, below, and then left alone.
It is not an argument about peer accountability systems or the design of consequences for poor performance. The only claim made here is that safety and demanding standards can and should coexist.
Nor is it about delegating decisions, protocols for judgment under heat, or the gradual drift of standards as small misses become normal. Each of those depends on psychological safety; none of them is the same thing. What this essay owns is the climate itself: what it is, why it is not soft, why it matters for decision quality, and how leaders build it.
3. What Psychological Safety Actually Is
Psychological safety is a shared belief within a team that interpersonal risks will not be punished. The risks are ordinary ones: asking a question that might reveal ignorance, admitting a mistake, flagging a concern about someone else's work, proposing an idea that might be wrong, asking for help. In a psychologically safe team, people expect that taking those risks will not bring embarrassment, rejection or a quiet loss of standing.
Three features of that definition matter for owners.
First, it is shared and team-level. Safety is not a personality trait of a confident individual. It is a climate that members of the same team tend to perceive similarly, and it can differ sharply between teams in the same organisation — between the civil package and the electrical package, or between the owner's design team and its commissioning team.
Second, it governs micro-decisions. Team members make dozens of small speak-or-stay-quiet choices every week. Most are never visible. Each one is a tacit calculation: will I be hurt, embarrassed or criticised if I say this? The aggregate of those calculations decides what the team knows collectively.
Third, it is not comfort. A psychologically safe team is not a cosy one, nor a team without pressure. Hard work, difficult goals and frank feedback are all compatible with safety. What safety removes is the fear that candour itself will be held against the speaker.
It also helps to be precise about what psychological safety is not.
It is not niceness or friendship. Teams of friendly people can be very unsafe for anyone who wants to challenge a shared assumption, because challenge feels like a breach of the friendship.
It is not cohesiveness. Highly cohesive teams can actually disagree less, because members value the cohesion and protect it. Cohesion can therefore reduce the willingness to raise an unwelcome view — the reverse of safety.
It is not the same as trust. Trust is typically about whether one person will give another the benefit of the doubt, built over time between individuals. Psychological safety is about whether others will give me the benefit of the doubt when I take a risk right now, and it describes the group rather than a pair. The two are related; a team with trust between members often feels safer. But they can diverge, and an owner who invests only in relationships may still find that people stay quiet in the formal settings where decisions are made.
A large internal study of team effectiveness in a major technology company reached a similar practical conclusion: of the dynamics it examined, psychological safety mattered most, and how a team worked together mattered more than who was on it. That finding comes from a corporate sample and should be read with that qualification, but it is consistent with the field research on which the rest of this essay draws.
4. Safety and Standards Are Separate Dials
The heart of the "soft management" objection is an assumed trade-off: more safety must mean less accountability. The research record rejects that trade-off directly. Psychological safety and accountability are separate dimensions, and treating them as opposites is a false choice. Skilled leaders reward excellence, sanction poor performance, and still welcome the errors, questions and bad news that uncertain work makes inevitable.
The phrase that captures the target is safety without complacency. Psychological safety means that no one is punished or humiliated for raising errors, questions or requests for help — and it means this in service of ambitious goals. Without the goal, safety drifts into comfort. Without the safety, the goal drives information underground.
Crossing the two dials produces four recognisable climates.
Low safety, low standards: disengaged. Expectations are modest and candour is risky. People do what is required and little more. Little is learned because little is attempted and less is said. On a capital project this looks like compliance-driven package management: forms completed, nothing volunteered.
High safety, low standards: comfortable and complacent. The team is pleasant and open, but nothing much is asked of it. Problems can be raised, yet there is no pressing reason to raise them, because performance is not demanding. This is the quadrant critics have in mind when they hear "psychological safety" — and it is not what the concept describes.
Low safety, high standards: anxious and silent. Expectations are demanding and the cost of candour is high. People work hard, but they also learn to manage how problems appear. Bad news is softened, delayed or routed around the leader. This is the common default on large capital projects: ambitious baselines, tight governance, aggressive commercial positions, and a team that has learned to report green until it cannot.
High safety, high standards: candid and learning. Expectations are demanding and the cost of candour is low. People raise problems early because doing so is expected and valued, and because the standard to which everyone is held makes early correction the rational choice. This is the quadrant owners should be aiming for.
The quadrant below is an interpretation built from the research on the false trade-off between safety and accountability. Its quadrant names are descriptive labels chosen for owner teams, not a reproduction of any published framework.
Figure 1: Psychological safety and performance standards are separate dials. "Soft management" is the comfortable, complacent quadrant; the common project default is anxious and silent; the target is candid and learning — safety without complacency.
Two practical consequences follow. First, an owner who moves a team from the anxious quadrant toward candour does not need to lower a single standard. The baseline, the gate criteria and the consequences for poor performance can all stay where they are. What changes is the reaction to bad news. Second, an owner who responds to the "soft" worry by raising pressure without touching safety will not get more performance. The team will get better at managing appearances, and the owner will see less of what is actually happening.
The related insight is structure without rigidity. High standards are not only targets; they include routines — reviews, after-action discussions, interface meetings — where reflection reliably follows action. Structure makes candour predictable rather than heroic.
5. Why It Matters: Learning, Decision Quality and Early Bad News
The mechanism linking safety to performance runs through learning behaviour. On complex, uncertain work, teams improve by asking for help, seeking feedback, discussing errors, experimenting and raising concerns. Each of those behaviours carries an interpersonal risk. Safety is the condition that determines whether people take the risk.
The field research frames this as a moderation effect. A clear, demanding goal gives learning a direction. Psychological safety determines whether that goal actually turns into learning behaviour. Without a goal, there is nothing to learn toward. Without safety, wholehearted learning feels too dangerous, and people fall back on protecting themselves.
Two lines of evidence support the mechanism. In a study of dozens of manufacturing teams in one company, teams with higher psychological safety showed more learning behaviour, and that learning behaviour was in turn associated with better team performance. In a study of hospital surgical teams adopting a demanding new cardiac technique, the teams that succeeded were those in which lower-status members — nurses, perfusionists, junior staff — felt able to speak up across a steep professional hierarchy, point out problems, and suggest changes during a difficult learning curve. Where speaking up felt risky, implementation stalled.
Neither study comes from construction, and owners should not overstate the transfer. But both describe the same essential situation that large capital projects face: interdependent specialists, uncertain technology or ground conditions, steep hierarchy, and a need to adapt as reality diverges from the plan.
The decision-quality consequence is direct. A decision can only be as good as the information that reaches the room. Low safety does not filter information randomly; it filters the inconvenient information first. Optimistic forecasts travel freely. Doubts about forecasts travel slowly. Early warnings about a supplier, a design interface or a ground condition are softened at each level until they reach the steering group as a minor watch item. The decision makers may be rigorous, independent and well informed about everything they are shown — and still decide on a cleaned-up picture.
Figure 2: A learning goal gives direction; psychological safety turns it into learning behaviour — asking, admitting, flagging, experimenting — so early bad news reaches the decision room. Low safety filters information first, and decisions are made on a cleaned-up picture.
This is why psychological safety belongs in a discussion of governance, not only of team culture. Gate reviews, independent assurance and risk registers all depend on the same raw material: what project people are willing to say. An owner can invest heavily in decision processes and still be poorly informed if the climate makes candour expensive.
6. Where Low Safety Costs Most on Large Capital Projects
The research above was conducted outside construction. What follows is an application by inference: the mechanism transferred to the structures typical of large capital projects. It identifies where the cost of low safety is likely to be highest, not where it has been measured.
Multi-employer hierarchies. A large capital project is rarely one team. It is an owner organisation, a project management contractor, several main contractors, many subcontractors and suppliers, and sometimes joint-venture partners — each with its own employer, career incentives and commercial position. A site engineer may report formally to one company, informally to another, and depend commercially on a third. Every additional layer adds a reason to stay quiet: speaking up may help the project while harming the speaker's employer. Safety at the level of the integrated team therefore has to be deliberately built; it will not arise from good intentions within each company.
Claim and liquidated-damages climates. When the commercial relationship is dominated by claims, notices and the threat of damages, every admission of error becomes potential evidence. The rational response is to withhold, which recreates exactly the withholding pattern seen when contractors in other sectors hid troubling test data from a punitive customer. The contract may be correctly drafted; the climate it produces still determines what the owner learns and when.
Safety and quality reporting. Near-miss reporting and nonconformance reporting are the clearest operational test of psychological safety on a site. A low reporting rate is rarely evidence of a safe site; more often it is evidence of a climate in which reporting carries a cost. The same logic applies to quality: nonconformances that are found and fixed quietly do not feed the learning that would stop the next one.
Front-end and interface work. The most valuable doubts on a capital project arise before sanction — in estimating, in design basis decisions, in early interface definition. These are also the moments when the sponsor's commitment to the preferred option is strongest and the pressure to show a fundable case is greatest. Estimators and designers who hold doubts at that stage face the steepest price for candour, and their doubts are the most expensive to surface later.
A counter-example. Large programmes have also shown the opposite pattern. On at least one major airport terminal programme, the client explicitly backed a speak-up culture across its integrated team, treating early disclosure of problems as expected behaviour rather than failure. That case is supporting illustration rather than proof, but it shows that a demanding owner can make candour part of the standard rather than an exception to it.
The common thread is that large capital projects combine high stakes, many employers, long durations and adversarial contracting — conditions that push the price of candour up. Owners who want early bad news have to push it back down deliberately.
7. How Leaders Build It: Frame, Invite, Respond
Psychological safety is shaped most strongly by leader behaviour. Team members watch closely how a leader reacts to the first piece of unwelcome news, the first admitted mistake, the first challenge to the leader's own view. That reaction tells everyone what candour will cost from then on.
The research identifies a set of leader behaviours that raise safety: being accessible, acknowledging one's own fallibility, explicitly asking for input from lower-status members, managing the effects of power, and framing new work as a challenging shared goal. A large corporate study of team effectiveness distilled a similar short list: frame work as a learning problem, acknowledge fallibility, and model curiosity. Brought together for owner teams, those findings fall naturally into three moves. The grouping below is a synthesis for practical use, not a reproduction of any single published toolkit.
Frame the work. Leaders set the conditions for candour before any problem appears. That means describing the project honestly as uncertain and interdependent — a job where no one, including the leader, can see every risk — and stating ambitious goals and clear expectations at the same time. A project framed as a routine execution of a settled plan makes every problem look like someone's failure. A project framed as demanding work under uncertainty makes early problem-finding part of the job.
Invite input. Framing is necessary but not sufficient; people still need to be asked. Leaders who are accessible, who say plainly that they will miss things and need to hear about them, and who explicitly ask lower-status members for their view generate far more information than leaders who simply declare an open door. Managing power runs in both directions: lifting the quiet voices and restraining the dominant ones, including the leader's own. Reflective sessions held without the immediate pressure of a decision — interface reviews, after-action discussions, short mistake reviews — create predictable occasions for candour. Some leaders make a habit of thinking out loud, exposing their own reasoning so that others can test it.
Respond productively. The response is where safety is won or lost. The minimum is to thank the person who brings bad news, whatever the content. Beyond that, leaders separate honest error from poor performance: honest mistakes on uncertain work become learning, while sustained underperformance and deliberate breaches still meet firm consequences. Responding productively also means closing the loop — showing what was done with a concern, so that raising the next one feels worthwhile. Some teams adopt explicit agreements on how mistakes and dissent will be handled, so that the expected response is written down rather than left to mood.
Figure 3: Frame the work as uncertain and ambitious, invite input by acknowledging fallibility and managing power, and respond productively by thanking the messenger, learning from honest error and still holding the standard — repeated at every meeting and every gate.
The three moves form a loop rather than a sequence. Each productive response reinforces the frame and makes the next invitation more credible. Each punished messenger breaks the loop, and one visible break can outweigh months of good framing. That asymmetry is why leaders on large capital projects need to be especially deliberate in the moments of highest pressure — the bad monthly report, the failed test, the surprise claim — because those are the moments the team remembers.
Owners can also build structure that makes these moves routine rather than personal. Gate criteria can ask for evidence that dissenting views were heard. Interface and design reviews can open by asking what has changed or gone wrong since the last review. Project-level agreements across the owner, the management contractor and the main contractors can state that early disclosure of problems will not, by itself, be used as commercial evidence. None of these replace leader behaviour, but they make candour less dependent on any one leader's temperament.
8. Traps That Make It Look Soft
Psychological safety earns its soft reputation mostly through poor implementation. Several traps recur.
Equating safety with removing consequences. "Anything goes" is not safety. It is the complacent quadrant, and it undermines the standard that gives candour its purpose. Safety removes the penalty for raising problems, not the penalty for causing them through poor work.
Treating it as a wellbeing initiative. Safety matters for people's experience at work, but on large capital projects its main value is as a performance and risk control: the mechanism by which early warnings reach decisions. Positioning it only under wellbeing invites the soft reading and lets project leadership treat it as someone else's programme.
Surveys and posters without changed responses. Climate surveys and value statements are cheap. Changed leader reactions to bad news are not. A programme that measures safety while leaders keep responding as before will teach the team that the programme is decoration.
Confusing a friendly team with a safe one. A cohesive, sociable team can be deeply unsafe for anyone who wants to challenge its shared view. Good relationships are welcome but are not evidence of candour in the room where decisions are made.
Asking safety to compensate for a non-team. Safety cannot rescue a group that lacks the basic conditions of a real team: a clear shared task, stable and clear membership, and a compelling direction. Where those are missing, the right fix is structural, and working on climate alone will disappoint.
Lowering standards to buy comfort. Some leaders, wanting to appear supportive, quietly relax targets or tolerate persistent underperformance. That moves the team toward complacency, and it confirms the critics' view that safety is soft.
Punishing the first messenger. The most damaging trap is to proclaim openness and then react badly to the first unwelcome news. The team learns more from that single reaction than from any amount of messaging, and it learns it quickly.
9. Close: Make Candour Cheap and Standards High
Psychological safety is not a concession to fragility. It is a hard-edged management discipline: the condition under which demanding standards produce truth rather than concealment. On large capital projects, where the most valuable information often sits with the people least able to afford raising it, that condition decides how early problems are found and how good decisions can be.
The owner's task can be stated simply. Keep the standards high — the baseline, the gate criteria, the consequences for poor performance. Make candour cheap — frame the work honestly, ask for bad news, thank the messenger, and show what was done with it. Neighbouring practices such as productive debate, trust, peer accountability, delegation and better meeting norms all depend on this climate; none of them replaces it.
Figure 4: Psychological safety is the climate beneath neighbouring ideas — productive conflict, silence misread as consent, rank and power, trust, accountability, standard drift and judgment under pressure. Each depends on it; none replaces it.
The test for any owner is not whether the team feels comfortable. It is whether the person who knows the bad news would tell the leader today — and whether the leader would still hold the standard after hearing it.
References
- Edmondson, A. C. (2002). Managing the risk of learning: Psychological safety in work teams. Harvard Business School working paper; in M. West (Ed.), International Handbook of Organizational Teamwork.
- Google re:Work. Project Aristotle — "Understand team effectiveness."
- Pellerin, C. J., Jr. (2009). How NASA Builds Teams. Wiley.
- Flyvbjerg, B., & Gardner, D. (2023). How Big Things Get Done. Currency.
- Lencioni, P. (2002). The Five Dysfunctions of a Team. Jossey-Bass / Wiley.
- Marquet, L. D. (2012). Turn the Ship Around! Portfolio.
- Shapiro, M. (2015). HBR Guide to Leading Teams. Harvard Business Review Press.
- Hackman, J. R., & Wageman, R. — conditions for team effectiveness.