Talk to the Person First

A team member becomes frustrated with a peer. The grievance itself is ordinary: a decision made without consultation, a commitment that slipped without warning, a meeting where the leader felt talked over. The first conversation about it happens with a trusted colleague, who listens and validates the interpretation. The next happens with a manager, framed carefully as a request for advice. Over the following weeks the concern travels through several more conversations, each one adding an example, sharpening the story, and quietly enlarging the audience. The one person who has not heard any of it is the peer at the center.

Most of the eventual damage is manufactured during those weeks. Validation from allies hardens an interpretation into a narrative, and the narrative starts collecting evidence, because a story in search of examples will find them. The peer, meanwhile, senses that something has changed. Meetings feel cooler, information arrives later, and nobody says why. They respond the way most people respond to unexplained distance, with caution and guardedness, and that response is then read as confirmation of the original concern. By the time the two people finally speak directly, if they ever do, the conflict is no longer about the decision or the meeting. It carries reputation, loyalty, an audience, and competing versions of history. An issue that might have been resolved in fifteen minutes of direct conversation has become organizational, and it became organizational through avoidance.

Across the operating-model and delivery organizations I have led and worked alongside, this pattern has repeated regardless of industry, seniority, or how capable the people involved were. Most workplace conflict does not become unmanageable because of the original disagreement, which is usually small, factual, and solvable. It becomes unmanageable through delay, through triangulation, and through follow-through so ambiguous that even a good conversation dissolves into competing memories within a month. The response I trust has two disciplines at its center. The first is old advice that organizations honor mostly in the breach: if you have a problem with someone, that person should usually be the first person you talk to about it. The second is less familiar: before any substantive conflict conversation, write the follow-up you expect to send afterward, and treat both the writing and the revising of it as part of managing the conflict itself.

The first conversation belongs to the person involved

The rule earns its place through what it protects. Talking to the person first gives them the chance to explain before anyone acts on incomplete information, and in my experience a meaningful share of conflicts end right there, because the missing context turns out to be real: the constraint nobody mentioned, the instruction from elsewhere, the message that never arrived. It keeps a private concern from becoming a public reputation, which matters because reputations assembled in a person’s absence are nearly impossible for them to correct. It reaches the disagreement while positions are still soft enough to move. And it establishes direct communication as the working norm, which is the condition that keeps managers from spending their careers as intermediaries between people who are perfectly capable of talking to each other.

The rule also implies a discipline for the leader on the receiving end of complaints. When someone brings a concern about a colleague, the most useful response is usually a set of questions: have you discussed this with them directly, what did you say, what stopped you, and would it help to prepare for that conversation together. The leader should support the conversation rather than absorb it. Absorbing it feels supportive in the moment and works against everyone involved, because it converts the manager into a messenger, denies the other person any chance to respond, and quietly builds a record the subject does not know exists. The same caution applies to confidentiality. A manager should not promise complete confidentiality before understanding the issue, because some concerns require action. A manager should also be wary of accepting a complaint on the condition that it never be raised with the person involved. Agreeing to that condition means agreeing to maintain a private grievance file, which serves nobody, including the person who brought the complaint.

I have watched this discipline do its work. One strong technologist I was developing into a manager brought me every problem he had with another person, and my first question was the same every time: have you talked to them? The repetition became a running joke between us, because his openings changed. He began arriving with the conversation already held, telling me he had a problem, had talked to the person about it, and thought he had solved it. The question he brought me changed too, from asking me to carry the conflict to asking whether he could have handled it better. The question I kept asking was boring, and that was its value. It refused, every time, to let the conflict come to rest with me, and he built his conflict skills in the only place they can be built, inside the direct conversations themselves. By the end he was a confident leader for whom resolving conflict directly was routine, and my contribution was mostly the discipline of never absorbing what belonged to him.

The rule needs its boundary stated plainly. Harassment, discrimination, threats, retaliation risk, serious ethical or legal concerns, and situations with a significant power imbalance do not belong in a talk-to-them-first frame. Those go to the appropriate management, human-resources, legal, or compliance channel, and no one should be expected to personally confront behavior that has made them unsafe. The rule governs ordinary working conflict: the disagreements, missed expectations, communication breakdowns, and strained working relationships where the overwhelming majority of organizational conflict actually lives.

Separate the facts from the story about the facts

When a conflict does need mediation, the leader’s role is narrower than most leaders assume. It is to establish what happened and which expectation applied. Deciding who has the more compelling story sits explicitly outside it, because stories are what the triangulation weeks were busy building, and rewarding the better one teaches the organization that narrative construction is how conflicts are won. The questions that do the work are unglamorous: what happened, when, who was present, what was said or done, what expectation applied and whether it was ever made explicit, whether this is one event or a pattern, and what remains genuinely uncertain.

Most of that work consists of sorting three categories that conflict tends to collapse into one. “I felt dismissed in the meeting” is an account of impact, and it is valid as exactly that. “You dismissed me because you do not respect my expertise” is an interpretation, and it assigns an intent that nobody in the room can verify. “You interrupted me three times before I finished explaining the delivery risk” is observable behavior, and it gives both people something they can examine together. Conflicts escalate when assumptions about intent get handled as if they were facts, and they de-escalate when someone insists on the difference. Emotion belongs in the same sorting. The anger or the sense of betrayal is real, it usually explains why the conflict escalated, and it deserves acknowledgment. It cannot serve as evidence of what happened, and it cannot be allowed to decide who is right or what the organization does next. The mediator’s recurring move is conversion: from emotional claim to the specific behavior, the context it occurred in, the impact it had, and the expectation that should govern it from now on.

Establishing facts sometimes means saying out loud that the facts are uneven. Mediation does not require pretending both accounts are equally supported. When the written record contradicts one version, when a commitment was explicit and missed, when one person owned a decision and did not make it, the leader should say so calmly, because balance manufactured to keep the room comfortable teaches everyone present that evidence is negotiable. Directness carries its own discipline: it applies to behavior and never to character. “You are not collaborative” is a verdict about identity; nobody can act on it, and most people will spend their remaining energy disputing it. “You finalized the staffing plan without consulting the two functional leaders whose teams would do the work” describes something that happened and can change. “You are defensive” invites more defense; “when the delivery concerns were raised, you questioned each person’s competence rather than addressing the examples they gave” identifies the behavior that has to stop. Specificity creates the possibility of correction, and the same rule governs whatever gets written down afterward: behavior, expectation, impact, and agreed action, never a character label that will outlive the conflict.

Write the follow-up before the conversation

Most leaders enter conflict conversations underprepared and leave them undocumented, and the two failures reinforce each other. The discipline that addresses both is simple to state: before the meeting, draft the email you expect to send after it. The value is in what the drafting forces. To write that email, the leader has to state the issue in neutral terms, list the facts actually established, separate them from the interpretations that still need testing, name the expectation that applies and whether it was ever explicit, identify the questions the conversation must answer, and describe what a usable outcome would include. A leader who cannot write that email in plain terms does not yet understand the conflict well enough to mediate it. The meeting that follows will be improvised, emotional, and difficult to summarize afterward, which is how apparently resolved conversations become the next round’s disputed history.

Drafting also exposes weaknesses while they are still cheap to fix. General impressions appear where specific examples should be. An intent shows up in the text that no evidence supports. An expectation the leader assumed was obvious turns out never to have been stated, which changes the conversation from accountability to clarification. A consequence appears that the established facts cannot yet carry. Every one of those discoveries costs a few minutes at a desk and would have cost real damage in the room.

The example that follows is deliberately hypothetical. A leader preparing to address a staffing decision made without the required consultation might draft:

I am summarizing today’s discussion so we have a shared understanding of the issue and what is expected next. We discussed the staffing decision made on Tuesday without review by the two functional leaders responsible for supplying the team. Based on what I currently know, the required consultation did not happen. Going forward, staffing changes that affect another function must be reviewed with that function’s leader before they are communicated as final. Specific actions, owners, and timing will be confirmed from our discussion.

The obvious objection is that preparing the outcome in writing biases the meeting toward that outcome, and the objection deserves a direct answer because the risk is real. The draft is a hypothesis, and the conversation is the test it has to survive. The leader must enter prepared to correct factual errors, add missing context, change the assessment, record disagreement accurately, and remove any conclusion the evidence no longer supports. The proof is in what happens afterward: sending the pre-meeting draft unchanged demonstrates that the conversation did not matter, and participants who compare the record against the meeting they actually attended will reasonably conclude the verdict was written in advance. Preparation makes the conversation sharper; refusing to let the conversation change the draft makes it theater, and the difference between the two is whether the record can still move.

Suppose the conversation surfaces something the leader did not know: the project deadline moved late on Monday, and the decision-maker believed immediate action was required. The revised version absorbs the new context without erasing the unmet expectation:

Thank you for the direct conversation today. We confirmed that the deadline changed late Monday and that Jordan believed immediate action was required. We also confirmed that the consultation requirement still applied and that no attempt was made to reach Casey before the decision was communicated. The staffing decision stands, and so does the expectation: changes affecting another function are reviewed with that function’s leader first, and if the matter is urgent and that leader cannot be reached within four business hours, it comes to me. Jordan will share the current staffing assumptions by Thursday, Casey will flag any capacity conflicts by Friday, and we will review whether this arrangement is working in two weeks.

The revision shows the conversation mattered. The new fact is recorded, the expectation survives it, the commitments have owners and dates, and the escalation path for the next urgent case is defined, so the same collision does not have to be relitigated. Timing does the rest of the work. A follow-up sent while the conversation is fresh lands as clarification; the same message sent five days later arrives after memories have diverged and the informal narratives have resumed. Prewriting is what makes promptness achievable, because the structure already exists and only what the conversation changed needs rewriting.

The record and the system

The final summary is a working document, and it should read like one: factual, concise, specific about expectations, and accurate about disagreement. It captures what was established and what remains disputed, with disputed points recorded as disputed rather than forced into false agreement, along with the commitments each person made, the support the organization will provide, when progress will be reviewed, and what happens if the issue continues. It goes to the people involved. Transparent documentation is a clarity mechanism first and a record second, and the ordering matters: a manager who keeps a private file while telling the participants the matter is informal has created hidden judgment, which is a worse condition than the original conflict. The choice is to document openly or not at all. The instinct here is the same one behind keeping decision records anywhere in an operating model: memory is a poor system of record, and a dispute about what was agreed is a conflict waiting to restart.

Some conflicts will not resolve through mutual understanding, and a leader who keeps prescribing better communication for them is avoiding a decision. When the real problem is unclear authority, conflicting goals, or accountability that was never evenly assigned, the parties cannot fix it by talking, because neither of them owns the fix. The leader decides, and the decision goes into the record. A different limit appears when one party will not engage in good faith: agreeing in the room and continuing the behavior, or disputing every written summary without ever offering a specific correction. At that point the issue has stopped being ordinary conflict and become a performance or conduct matter, and the record of expectations already communicated and support already provided is what allows the organization to treat it as one instead of starting over from memory.

A conflict that keeps recurring is often less a statement about two people than a symptom of the operating model they share. Overlapping authority, unclear decision rights, incentives that reward local optimization, and scarce resources that force the same competition every quarter will produce the same collision no matter who holds the roles. The question worth asking at the end of any mediation is what about the system made this conflict likely, because replacing a person while leaving the condition intact tends to reproduce the conflict with a new cast. When the resolution includes a structural change, such as a clarified decision right, a consultation step, or an escalation threshold, that change belongs in the written record alongside the personal commitments.

None of this aims at an organization without disagreement. Disagreement about priorities, risk, and resources is evidence that people are engaged with real trade-offs, and its complete absence usually signals fear or apathy rather than alignment. The aim is narrower: to keep disagreement from turning personal, political, or permanent. Success does not require that everyone agree about the past. It looks like this instead: the people involved understand what happened and what is expected, the commitments have owners and dates, the work continues, and if the conflict returns, the organization can tell the difference between a misunderstanding and a failure to meet an expectation that was clearly set. Direct conversation makes that outcome possible. The prepared, revised, and promptly sent record is what makes it durable.