Leadership ethics: how decisions reveal values

Nobody in your organization forms a view of your leadership ethics by reading what you wrote. They form it by watching what happened to the person who missed the number, and what happened to the person who hit it in a way everyone knew was ugly.

That gap, between the values a company states and the values it reveals under pressure, is the whole subject. Everything below is about closing it, and about the fact that closing it costs money, people, and sometimes your own standing.

What to take away

Most ethical failures at leadership level are not decisions to do wrong.
Information about problems degrades as it moves up.
Leadership writing tends to imply that the ethical option is also the profitable one over the long run.

Stated values and revealed values

A stated value is what appears on the wall. A revealed value is what you can infer from a sequence of decisions. When the two conflict, people believe the second one, and they are right to.

The revealed value is visible in a small number of places:

Who got promoted.If a manager with a reputation for bullying advances anyway, the organization has decided that results outweigh conduct. It has decided this whether or not anyone said so out loud.
Which exceptions were made.A rule that is waived for the top of the sales table is not a rule. It is a tax on the people without bargaining power.
What happened after a complaint.Not the investigation's finding: the finding is usually private. What people see is whether the person complained about is still there, still in the same seat, still their manager.
Which questions get a real answer.If a direct question in a town hall gets a rehearsed non-answer, people learn the topic is closed.

If you want to know what your organization actually believes, do not survey it. Read the last two years of promotion decisions and disciplinary outcomes side by side.

The top-performer test

Most ethical failures at leadership level are not decisions to do wrong. They are decisions not to act on someone who is producing.

The pattern is worth naming. Someone brings in a large share of revenue, holds a critical technical system in their head, or has a customer relationship nobody else has. Their conduct is a known problem, and every level below you has tried to raise it.

The cost of removing them is immediate, quantified, and lands on you, while the cost of keeping them is diffuse, delayed, and lands on other people.

So they stay. And the whole organization learns the exchange rate: this much production buys this much misconduct.

What to do about it is unpleasant but not complicated:

Work out what removing the person would actually cost, in cash and in months, rather than accepting the estimate from the person most invested in keeping them.
Notice who is producing the estimate. The manager who benefits from the performance is the wrong person to size the risk.
Ask what you would do if the same conduct came from someone in the bottom quartile. If the answer is different, you have your answer about what your standard really is.
If you decide to keep them anyway, sometimes there is a genuine reason, say so explicitly to the people affected, and say what you are doing to protect them. An acknowledged compromise damages less than a pretended standard.

The version of this that does the most harm is the quiet transfer: the person is moved sideways, the problem moves with them, and the receiving team is not told. Everyone who knows watches you choose the organization's comfort over the next team's safety.

Incentives beat policies, every time

You can write a policy in an afternoon. You cannot make it beat a compensation plan.

If a target is aggressive enough that hitting it honestly is rare, people will find another way to hit it, and most of them will not experience that as a moral choice. It will feel like ordinary resourcefulness. The pressure arrives as a number and a deadline; the ethics arrive as a document nobody has opened since onboarding.

Before you set a target, ask three questions:

Incentives beat policies

Can this be hit honestly by an average performer in an average quarter?If only the exceptional can hit it cleanly, the average will hit it dirtily.
What is the cheapest way to hit it that I would not want to read about?Whatever that is, someone will find it. Design the measurement so it does not work.
What happens to someone who misses and says why?If the honest miss is punished as hard as the dishonest hit, you have priced honesty out of the market.

The same logic applies to anything measured: support-ticket close rates that reward closing over resolving, utilisation targets that reward billing over usefulness, safety-incident counts that reward not reporting.

A useful discipline: when you add a metric with consequences, write down how it could be gamed. Then add a second measurement to catch that. If you cannot imagine the gaming, you have not thought long enough.

The same preference for evidence that a program changes behavior, not that it exists, runs through the U.S. Department of Justice Criminal Division’s Evaluation of Corporate Compliance Programs (ECCP).

What people read the signal from

The statement	The signal people actually read
"We have an open-door policy"	Whether the last person who used it is still progressing
"Safety comes first"	Whether the line gets stopped when stopping it costs a shipment
"We compete on the merits"	What happened to the deal that could only be won one way
"Speak up, you'll be protected"	Whether anyone can name a case where speaking up worked out
"We hold everyone to the same standard"	Whether the standard survived contact with a big producer
"Bad news early"	What happened to the last person who brought bad news early

None of these signals is under your rhetorical control. All of them are under your decision-making control.

Stated value vs signal read

The statement

Open-door policy: Is the last user progressing?
Safety comes first: Is the line stopped at cost?
Compete on merits: What happened to that deal?
Speak up, protected: Can anyone name a case?
Same standard for all: Did it survive a big producer?
Bad news early: What happened to the last bringer?

The signal people read

Open-door policy
Safety comes first
Compete on merits
Speak up, protected
Same standard for all
Bad news early

Making yourself the last to know

Information about problems degrades as it moves up. Each layer has a rational reason to soften it: the person who reports a problem often becomes the person who owns it, and the person who owns it is the person blamed for it.

Leaders usually respond to this by asking people to be more forthcoming. That does not work, because the request does not change the incentive. What changes the incentive:

Separate the messenger from the owner.If reporting a problem automatically makes you responsible for fixing it, reporting drops. Make the assignment a separate decision.
Ask for the specific thing, not the general thing."Anything I should know?" gets nothing. "What is the thing most likely to be in the press in six months, and who told you about it first?" gets more.
React visibly well once.People calibrate on cases they can name. One well-handled piece of bad news, discussed openly, does more than a year of encouragement.
Go around the layer occasionally.Skip-level conversations are uncomfortable for the layer skipped, and that discomfort is part of the price. Tell the layer you are doing it and why.
Notice silence.A function that never reports problems is not a function without problems.

Three situations, worked through

These are hypothetical. They are constructed to show where the conflict actually sits, not to describe anything that happened.

A deal that needs a favor. A large public-sector contract is close. The intermediary who introduced you asks for a success fee well above the market rate, and is vague about what the work consists of. Nothing has been alleged and no rule has obviously been broken.

The conflict is not "should we pay bribes", nobody frames it that way internally. It is between a large certain revenue and an ambiguous cost. What makes it resolvable is refusing the ambiguity: ask what the fee buys, in writing, in specifics.

The Foreign Corrupt Practices Act (FCPA) and the UK Bribery Act are examples of laws that may apply. Anti-corruption rules differ sharply by country and by the sector you sell into, and the personal exposure is not always the company's; this is a question for a lawyer before it is a question for a leadership team.

A defect that is probably fine. Testing turns up a fault that appears in a small fraction of units and is unlikely to injure anyone. A recall would be expensive and public. Delaying means more units in the field.

The tempting move is to keep gathering data, because more data is always defensible. Notice when "we need more information" has become a way of not deciding. Set the threshold in advance: what result would make us act, and by when do we need it?

Write that down before the next test comes back, while you still do not know which way it will go. Product safety and recall duties are legally governed; for U.S. consumer products, the Consumer Product Safety Commission (CPSC) is a key federal recall authority, and the Consumer Product Safety Act can require reporting of potential hazards. Duties vary by market, and counsel and the regulator, not the leadership team alone, determine what must happen and when.

A layoff you can do two ways. You need to reduce headcount. One approach gives people long notice and support, and costs more. The other is faster and cheaper, and people find out on the day.

There is no version of this where nobody is hurt, and pretending otherwise is its own failure. The ethical content is in the parts you control. Were people told the truth about why? Were the criteria for selection defensible and applied consistently?

Did managers deliver the news themselves? Did you say what you were doing rather than describing it as a restructuring? For covered U.S. plant closings and mass layoffs, the federal Worker Adjustment and Retraining Notification (WARN) Act requires advance notice; state laws may add obligations. Employment obligations around notice, consultation, and selection differ substantially by jurisdiction, so get advice on what is required before deciding what is generous.

When two obligations genuinely conflict

Some decisions are not tests of character. They are conflicts between real duties, where every option breaks something.

Closing a site keeps the company solvent and removes a town's largest employer. Honoring a supply contract to a customer means breaking a promised delivery to another. Disclosing a problem early protects users and hands an advantage to a competitor who has the same problem and is saying nothing.

For these, a decision procedure is worth more than a principle:

Name both obligations out loud,to the people in the room. Half the damage in these decisions comes from pretending only one obligation exists.
Identify who bears the cost of each option,by name or by group. Options whose costs land on people not in the room get chosen too often.
Look for the option that keeps part of both.It is usually worse on paper than either pure option and better in practice.
Decide who decides,and at what level. A decision that should have gone up and did not is a governance failure regardless of how it turned out.
Write down the reasoning while you are uncertain.A record made after the outcome is known is a justification, not a record. The version written in advance is the one that helps the next person.

The costs, stated plainly

Leadership writing tends to imply that the ethical option is also the profitable one over the long run. Sometimes it is. Often the timescales do not line up: the cost is this quarter and yours, the benefit is in three years and someone else's.

Be honest about that internally, because people can tell. Saying "this will cost us and we are doing it anyway" is credible. Saying "doing the right thing always pays" invites everyone to notice the cases where it did not, and to discount the whole message.

The costs, stated plainly

Revenue you decline, invisible in competitor numbers
Large time spent on unmarketable investigations
Political capital spent overruling a right senior person
Occasionally, your own position

None of this is an argument against doing it. It is an argument for not being surprised, and for not promising your organization a painless version.

When you got it wrong

You will. The question is what happens next, and this is where most of the remaining damage is done.

The default corporate response (minimise, delay, characterise it as a process issue, let it go quiet), is usually worse than the original problem, and it is the part people remember. It also teaches everyone below you how to handle their own mistakes.

What works better is unglamorous: say what happened, what it cost, and to whom, and say what you are changing. Accept that the first telling will be worse for you than the fifth.

If people were harmed, address the harm before the narrative. If someone raised it earlier and was ignored, say that too, the single most credible thing you can say and the single most tempting thing to leave out.

Questions worth asking yourself

Which of my decisions in the last year would I not want explained accurately to the team affected?
Who in this organization is protected by their results, and who knows it?
When did I last hear bad news early enough to act on it, and what happened to the person who brought it?
Is there a target in this business that cannot be hit honestly?
If someone raised a serious concern about a peer of mine, what would actually happen to them?

If a question is hard to answer because you do not have the information, that is the finding. Go and get it before you write anything else about values.

Where a concern reaches you about conduct of certain kinds, the person raising it may also have protection in law rather than only in your goodwill. In the UK, the Public Interest Disclosure Act 1998 (PIDA) provides protections against detriment and unfair dismissal for qualifying workers who make protected disclosures; claims can be brought to an Employment Tribunal, as the UK's whistleblowing framework explains.

Related reading on this site: ethical frameworks covers the reasoning behind the decisions above, and codes of conduct covers the document your conduct either backs or undermines.

Common questions

I inherited a team with bad habits. Where do I start?

Find out what is currently tolerated before you announce anything. Ask each person what the last manager let slide and what the team does that they would not want described. Then name three things you will not accept, and hold the first test.

How do I handle a top performer whose conduct is a problem?

The same week, privately, with the rule stated and the consequence named. Where the conduct is serious you are into a process governed by employment law and your organization's procedures, and the people function runs that rather than you.

What if the pressure comes from above me?

Ask what the request is for, who else approved it, and whether you can have it in writing. Those three questions are what a careful person asks, and a request that cannot survive them should not be carried out.