Values, integrity, and ethical judgment under pressure

A culture is the accumulated pattern of which behaviour gets rewarded, tolerated and punished, and a values poster is no part of it. What defines it is what leaders do when the stated values become costly. Everyone watches the gap between word and action, and everyone believes the action. v1.0 framed this competency as modelling values under pressure. v3.0 extends it to ethical judgment, because the situations that test a manager hardest are those where the organisation's stated values, the law, the policy and the pressure of the moment leave the question open and the manager has to settle it. The competency is upholding values at exactly the point where doing so is expensive. That means dealing with the brilliant jerk instead of banking their output, reporting status truthfully without shading it, and obeying the error budget when the launch is due. At scope it also means running culture as the system it is, made of what leadership visibly does, rewards, promotes and tolerates. It matters because tolerating a behaviour endorses it, because a value with an exemption for seniority has become a price, and because one visible exception undoes quarters of work. The examinations test it through the high performer whose behaviour buys silence, the manager asked to report a slipping project as on track, and the leadership that shipped through an exhausted error budget in full view of everyone.

1. The costly moment is the culture. Values that last only while they cost nothing are decoration. When the right call costs a launch, a customer or a person, what the leader does is what the organisation learns and copies.
2. Tolerating is endorsing. Behaviour that is rewarded gets repeated, and behaviour that is tolerated gets learned. Banking the brilliant jerk teaches everyone that results buy exemption.
3. Consequences reach the powerful, or they are pricing. A value enforced only below a certain level has stopped being a value. A senior, high-impact leader who breaches one faces real and visible consequences, and how results were achieved counts as part of the result.
4. Do not shade the truth upward, and absorb the cost of declining. When asked to report a status, a rating or a commitment as better than it is, the manager declines. The manager carries the cost of declining and does not pass it down to the team.
5. Behaviour changes when the system changes. A third mandate for something two mandates did not fix will fail. Behaviour shifts when the desired thing becomes the easy path, is visibly valued, and takes less effort to do. Culture change means realigning hiring, promotion, recognition and consequences over many quarters, and an announcement will never achieve it.
6. Where the rules do not settle it, judgment does — and it is explained. A request may be legal and wrong, a policy may produce a bad outcome in this particular case, or one value may pull against another. The manager decides on the merits, gives the reasons and owns the decision.

The research literature on culture is deep, and the experimental evidence on how to change it is thin. The unit grades its models as practice and takes its working definition from the founding work in the field.

Artifacts, espoused values, and assumptions practice

Edgar H. Schein, Organizational Culture and Leadership (1985) describes culture at three levels: the visible artifacts, the values the culture espouses, and the underlying assumptions that drive behaviour. Those assumptions are usually unspoken and often contradict the espoused values. This unit uses the model to diagnose. Where the espoused values and the behaviour diverge, the behaviour is the culture. The assumptions it exposes, such as "results buy exemption" or "the budget is obeyed unless it is inconvenient", are what a leader is in fact managing.

Culture as what is rewarded, tolerated, and punished practice

This is less a named model than the working definition this standard uses. Culture is the accumulated pattern of how people behave, and it is set far more by what leaders visibly do, reward, promote and tolerate than by stated values, policies or announcements. For change, it follows that culture is realigned through the systems of hiring, promotion, recognition and consequences. A single visible exception, such as the brilliant jerk who is tolerated or the value waived for a launch, can undo quarters of work.

Service-level objectives and error budgets practice

This is the mechanism from Betsy Beyer, Chris Jones, Jennifer Petoff and Niall Richard Murphy (editors), Site Reliability Engineering: How Google Runs Production Systems (2016), introduced in DE-3 Reliability, incident response, and operational ownership and treated here as the unit's defining test case. An error budget turns "reliability first" from a slogan on a poster into a rule agreed in advance, and it is worth only as much as leadership's willingness to obey it when a launch is due. Shipping through an exhausted budget in front of the whole organisation teaches everyone the real policy. Obeying it at a cost is the most legible cultural act an engineering leader can perform.

Leading change practice

SV-6 Organisational design and leading change uses the model from John P. Kotter, Leading Change (1996) for structure, and this unit uses it for norms. Credible leaders form a coalition and model the new behaviour, systems are realigned behind it, early wins are made visible, persistence is measured in quarters, and the change is anchored in how decisions are made. In culture change the steps most often skipped are the last ones, consolidation and anchoring. The announcement only starts the process.

The costly moment practice

The unit is organised around this concept, which is not a named model. It describes a situation where a stated value and the pressure of the moment point in different directions and people are watching the decision. It comes from experience, and because it is so easy to recognise, nearly every scenario in this competency is built from one.

The values audit, in behaviour

Once a year the manager lists the organisation's stated values and writes beside each one what happened the last time it was costly. Where the two diverge, the divergence is the culture, and the manager names it as such before changing anything.

The brilliant-jerk decision

When a high performer's behaviour breaches the team's values, the manager acts within the month. They give specific feedback, set an explicit expectation and a date, and let consequences follow through PL-4 Performance management if the pattern continues. Strong output is no reason to wait, since every week of tolerance announces the real values.

Declining to shade

When asked to report a project, a rating or a forecast as better than it is, the manager says no plainly, offers the honest picture with options, and bears the discomfort personally instead of passing it to the team. This habit is what makes the manager's good news believable.

Changing behaviour by changing the system

For a norm the organisation wants, whether documentation, blameless reviews or a practice spreading from one team to others, the manager makes it the easiest route to something people already want. They value it visibly in recognition and promotion and reduce the friction. Nobody issues a mandate for what two mandates failed to fix.

Consequences at the top

When a senior leader breaches a core value, the consequence is real and visible, and how they achieved their results is treated as part of their performance. The executive begins with the mirror. If the leadership team models behaviour the organisation should not copy, that is the first culture problem, and it is handled with the directness expected of any manager.

Owning the contradiction

Sometimes leadership itself breaches a stated value, by shipping through the budget or waiving the process for a launch, and everyone notices. The only repair is to own the contradiction publicly and behave differently at the next costly moment. Anything less turns the value into décor.

At team scope the object is the example set in the room: values upheld when they cost something, credit given, mistakes owned, the brilliant jerk dealt with. At organisational scope the manager learns how culture works. Behaviour is the culture, it is set by what managers do every day, it changes when the system changes, and practices spread by pull. The manager also enforces the values between people, for instance by backing a new manager against a senior engineer who undermines them. At executive scope culture is the executive's system. The costly moment is decided in line with the stated value, or the contradiction is owned. Consequences reach the powerful. Culture change runs as a realignment over many quarters. Well-loved but costly norms are evolved together with their champions. Culture is built into the institution so that it survives growth, and direction is carried by line leaders who visibly believe in it. The pattern is set out in How Judgment Scales; the safety that lets people say when a value has been breached is covered in CC-1 Psychological safety and productive conflict.

• Upholds values at exactly the moments they are costly, knowing the team is watching. team
• Gives credit, owns mistakes and respects boundaries, so that actions match stated principles. team
• Deals with a values breach by a top performer instead of banking the output. team
• Recognises when they are being asked to shade the truth about status, and declines.
• Failure mode: values that evaporate under deadline pressure. team
• Failure mode: saying one thing and rewarding another. team
• Failure mode: assuming you only model behaviour when you mean to. team
• Reads culture from behaviour instead of posters, and treats any divergence from the stated values as the real culture. org
• Changes behaviour by changing incentives and friction, and does not issue a third mandate for what two failed to fix. org
• Spreads a good practice by pull (visible value, easy adoption, champions) and avoids decree. org
• Deals with a respected senior engineer who quietly undermines a newly promoted manager directly and in private, calling the behaviour what it is, while backing the manager in full view. org
• Holds managers to the organisation's values, including the manager who keeps a toxic high performer because the numbers are good. org
• Declines to shade status, ratings or commitments upward, and absorbs the cost of declining instead of passing it down. org
• Failure mode: a third mandate for something two mandates did not fix. org
• Failure mode: expecting a values statement to outvote the reward system. org
• Failure mode: leaving a new manager to survive a senior saboteur alone. org
• Decides the costly moment in line with the stated value, for example by obeying the error budget with the launch due, or owns the contradiction publicly and behaves differently next time. exec
• Makes sure consequences reach the powerful: a senior leader who breaches a core value faces real and visible ones, and a director who hits every number through fear is acted on for how they did it. exec
• Starts with the mirror and fixes the behaviour the leadership team itself models first, because the organisation copies the top table. exec
• Runs culture change as a realignment of the system over many quarters, with credible leaders modelling it, incentives and promotions realigned, early wins made visible and the effort sustained, and treats the announcement as no more than the starting gun. exec
• Evolves a well-loved but costly norm without breaking trust, by honouring what it did well, involving its champions in what replaces it and being honest about what changed and why. exec
• Builds culture into the institution so that it survives growth beyond a few hundred people, with leaders hired and promoted for it, norms taught and systems aligned. exec
• Carries direction through line leaders who visibly believe it; if direction dies two levels down, it was announced and never communicated. exec
• Exercises ethical judgment where the law, the policy and the pressure leave the question open, and explains the decision. exec
• Failure mode: shipping through the error budget and expecting the poster to survive. exec
• Failure mode: values with an exemption for seniority. exec
• Failure mode: culture change by announcement at an all-hands. exec
• Failure mode: scaling by dilution. exec
• Failure mode: mistaking cascade emails for communication. exec

Results versus methods. An individual or a leader who delivers while breaching the values forces the question of what performance means. This standard's answer, as in PL-4 Performance management, is that the methods are part of the result. The pull towards the number is strong, and the examinations test whether the candidate resists it.

Consistency versus context. Values applied identically every time can produce a bad outcome in a case the rule did not anticipate, and values applied with judgment can look like values applied selectively. The resolution is to decide on the merits, give the reasons and own the decision. The explanation is what separates judgment from exemption.

Change versus trust. Changing an entrenched norm, even a costly one, breaks something people were attached to. Evolving it with its champions, honouring what it did well and being honest about the reasons takes longer than announcing a new norm, and it is the only route that does not use up trust.

Honesty upward versus safety downward. Refusing to shade a report costs the manager standing with their own leadership. Pushing the team to make the report come true costs the team. The manager takes the cost, and taking it is the competency.

Rules versus judgment. Some requests are legal and wrong, and some policies would cause harm in a particular case. The rule leaves these open and the manager's judgment has to close them. Judgment is harder to defend than compliance, so the manager explains the reasoning as carefully as the decision itself.

A head of engineering leads an organisation whose stated value, painted on the wall and repeated at every all-hands, is "reliability first". Six weeks ago a major launch fell due while the core service's error budget was exhausted. The organisation's agreed rule (budget exhausted, so stabilise) meant the launch should have waited. The chief executive wanted the date and the head of engineering agreed to ship. The launch went out, and a week later the core service had its worst incident of the year. Everyone in the organisation saw the sequence. Since then two teams have quietly stopped tracking their error budgets, and a principal engineer has said in a skip-level that "reliability first" has become a joke.

The cultural significance is total. What the organisation took away was the real value, that the budget is obeyed unless a launch is due, and the importance of the launch did not register. Two easy responses are available. One is to say nothing and let the incident fade. The other is to restate the value with a new policy and firmer wording. The first confirms the lesson. The second is a third mandate for something the organisation has just watched leadership break, and it will be read in exactly that light.

The only repair is to own the contradiction in public and act differently at the next costly moment, and the head of engineering does both. At the next all-hands they describe plainly what happened. The organisation had a rule, leadership broke it for a launch, the incident followed, and the two teams that stopped tracking their budgets drew the right conclusion from what they saw. They say the contradiction was their own, since the head of engineering agreed to ship, and that the value is worthless unless it holds when it is expensive. They do not explain why the launch mattered, because the organisation already understood that and it is beside the point.

Then comes the behaviour. The head of engineering reinstates the budget rule with one change. A decision to ship through an exhausted budget can only be taken at the executive table, in writing, with the reason and the accepted risk recorded and published to the organisation. The rule is tested four weeks later, when a second launch meets a red budget on a different service. The head of engineering holds the launch, at a visible cost to a customer commitment, and says so. The chief executive hears the reasons in advance, set out in terms of both the organisation and the business: the value exists only if it costs something, and the last time it was waived the organisation paid more.

The two teams start tracking their budgets again without being asked. The principal engineer says nothing, which is a signal in its own right. The head of engineering does not send round a reminder about the value. It would have been the cheapest response, and it would have finished the value off.

• PL-4 Performance management: the brilliant jerk, and the leader who delivers through fear, treated as performance problems.
• CC-1 Psychological safety and productive conflict: the safety that lets people say a value has been breached, and trust as a behavioural account.
• DE-3 Reliability, incident response, and operational ownership: the error budget as the mechanism, and the review after the failure.
• DE-2 Estimation, planning, and honest commitments: declining to give a false forecast under pressure.
• SV-6 Organisational design and leading change: leading change through structure, and evolving norms with their champions.

1. Why do teams believe actions over stated values?
   Answer

   Culture is the accumulated pattern of what is rewarded, tolerated and punished in practice, and the costly moment exposes it. A value that only holds while it costs nothing tells people nothing. What a leader does when it is expensive is what they learn and copy.
2. What does tolerating a brilliant jerk teach a team?
   Answer

   That results buy exemption from the values. Each week of tolerance announces the real culture more loudly than any statement. The fix is feedback, an expectation, a date and consequences, all within the month.
3. You are asked to report a slipping project as on track. What do you do, and who carries the cost?
   Answer

   Decline plainly, and give the honest picture with options. The manager bears the cost of declining and does not push the team to make the report come true.
4. Two documentation mandates failed. What does the third attempt have to change, and why? org
   Answer

   It has to change the system, and a new mandate will not do that. Documentation becomes the easiest route to something people already want (onboarding, review, incident response), writing it is visibly valued, and keeping it up takes less effort. Behaviour changes when the system changes, and a third mandate for what two did not fix will fail the same way.
5. A respected senior engineer is subtly undermining a newly promoted manager. What is the response? org
   Answer

   Speak to the engineer directly and in private, calling the behaviour what it is, while backing the manager in full view. Letting it pass teaches the department that anyone with tenure can negotiate with authority.
6. Leadership shipped through the error budget and everyone noticed. What did the organisation just learn, and what is the only repair? exec
   Answer

   It learned the real value: the budget is obeyed unless a launch is due. The only repair is to own the contradiction publicly and behave differently at the next costly moment. A reminder or a new policy turns the value into décor.
7. A director hits every number and leaves a trail of burned-out people and fear. Why must the executive act, and on what? exec
   Answer

   The executive acts on how the results were achieved. Every quarter of tolerance announces the organisation's real values, and a value enforced only below a certain level is a price and has stopped being a value. Consequences have to reach the powerful.
8. How does culture survive growth past a few hundred people? exec
   Answer

   Through deliberate institutionalisation, with leaders hired and promoted for it, norms taught and systems aligned. At that size the founders' pull no longer reaches everyone, and dilution is a responsibility for leadership to manage, not weather to be endured.

• Edgar H. Schein, Organizational Culture and Leadership (1985): artifacts, espoused values and underlying assumptions.
• Betsy Beyer, Chris Jones, Jennifer Petoff and Niall Richard Murphy (editors), Site Reliability Engineering: How Google Runs Production Systems (2016): error budgets as a rule agreed in advance, and what obeying them costs.
• John P. Kotter, Leading Change (1996): the steps of change, and the ones most often skipped.