How to talk after a team failure without looking for someone to blame

The outcome is known, but the earlier decisions are not

An order was not delivered on time. At the meeting, the director asks: “Who failed to keep an eye on the deadline?” The production manager points to logistics; logistics replies that it was never told about the change of plan. The team already knows the painful consequences, so the earlier decisions start to look like obvious mistakes. That is a trap: we judge the quality of those decisions on the basis of a later outcome that nobody knew at the time. Research on judging events in hindsight shows that after an event it is easy to see it as more predictable than it really was. That is why the facilitator starts differently: “What did we know on Tuesday at ten? What alternatives did we have? When did the first sign of the change appear?” Only questions like these make it possible to check whether the problem was a bad decision, a poorly designed process, a lack of the necessary resources or a deliberately ignored warning. They do not settle responsibility in advance.

A safe conversation does not mean leniency

Amy Edmondson’s research on teams showed that psychological safety is linked to a willingness to report problems and to learn. A later meta-analysis by Frazier and colleagues confirmed the links between psychological safety and behaviour at work, although it does not prove that a single conversation technique will automatically improve results. For a leader, the conclusion is practical: the person who noticed a mistake must be able to speak, even if they are lower in the hierarchy. At the same time, you need to distinguish an error in a complex process from withholding information, deliberately getting round safeguards or breaking the law. Saying “Nobody is being held to account” would be dishonest. A better version is: “In this meeting we’re going to understand how the incident unfolded and plan the fixes. If any issues come up that require a formal investigation, we’ll deal with them separately, in line with the rules.” This sets predictable boundaries for the conversation.

Review the incident, don’t put a person on trial

Prepare a simple timeline: the expected result, decision points, warning signs, handovers of information and the actual outcome. Everyone can add a fact and its source: a message, a system entry, notes from a briefing. Where there is no certainty, write “not established” rather than filling the gap with guesswork. In practice, four questions work well: what did we intend to do, what actually happened, why was there a difference, and what will we change next time? They are not a magic formula. In a meta-analysis of 46 samples involving 2,136 people, Tannenbaum and Cerasoli found an average improvement in effectiveness after structured debriefs. You must not, however, promise every company a specific percentage improvement: the studies covered a range of sectors, teams and simulations. The strength of a review lies in its structure, which keeps participants talking about the work rather than about their colleagues’ character traits.

When people get defensive, go back to observations

“But I warned you” and “You always dump everything on us” are understandable reactions, but they shift attention from the process to reputations. The facilitator can respond: “Let’s stay with this particular piece of information. When was it passed on, and to whom? Did the recipient confirm what they would do with it?” If one person dominates, it is worth asking participants in turn for a brief account, without letting others interrupt. If the conflict is acute, some of the facts can first be gathered individually. Before you assess someone’s decision, ask them to set out the information and constraints they were aware of at the time. This is not about excusing everything, but about avoiding a situation in which the version told by whoever speaks loudest is taken as the truth. If a new fact emerges that reflects badly on company procedures, the leader should accept it, even when it is uncomfortable for their own department. Otherwise the next warning signs may go unspoken.

From conclusions to fixes: every fix has someone responsible for it

A summary such as “We need to communicate better” is not corrective action. Replace it with a record of what exactly we are changing, who is responsible, by when, and how we will know the change is working. For example: when production changes an order deadline, the system automatically flags the change, and someone from logistics confirms receipt before the next briefing. The operations manager is responsible for this process, and the first review will take place after two weeks. If a correction requires a board decision, write down who will prepare the options and to whom they will present them. Distinguish responsibility for a past mistake from responsibility for future improvement. Someone may not have caused the failure and yet be the right person to change the instructions. Remember the side effects too: an extra form may improve documentation but slow down the response in an urgent situation. It is best to test the new rule on real examples.

Don’t end the meeting just because the tension has eased

An important result of the conversation is checking that people understand the decisions in the same way. Ask each person responsible for an agreed action to repeat briefly: “What will you do, by when, and who will you inform?” The record should distinguish established facts, hypotheses still to be tested and decisions already taken. Send it to people who were not at the meeting but will be involved in putting the changes into practice as well. After the agreed period, do not just ask whether the fix has been introduced; look at the cases in which it should have worked. Did people get the signal sooner? Has the gap in responsibility closed? Has the new process simply shifted the problem onto another shift? It is also worth coming back to the atmosphere in the team: is anyone afraid to report similar issues? If so, a separate conversation is needed about how leaders respond. Analysis of a failure ends only when the knowledge changes practice.

A conversation after a failure should help explain what happened and prevent it from happening again, not provide an easy scapegoat. A reliable timeline, space for uncomfortable facts and concrete corrective actions give a team more than a declaration that “we must learn our lessons”. Accountability still matters, but you need to know exactly what it relates to.

How Em helps you prepare a meeting after a failed project

Agnieszka heads a department that has missed an important deadline. Before the meeting, she describes to Em the facts available: the dates the order was handed over, changes to the schedule, the lack of confirmation from logistics and the tension between two people. She asks for help preparing a neutral opening and questions that will not prejudge who was at fault. Em helps her separate certain information from interpretations and suggests how a thirty-minute review could run. Agnieszka also practises her response to the sentence: “It’s not our fault – you always offload everything onto us.” Instead of arguing about intentions, she asks about the most recent decision and the document it was based on. If she knows the team’s collaboration preferences, she can prepare a way of gathering input that does not favour only the most assertive people. After the meeting, she uses Em to structure the follow-up message: what we know, what we are checking, which corrections we have agreed and when we will come back to them. A conversation prepared in this way gives employees the chance to tell the truth and gives the manager concrete commitments to follow up on.