Reporting a Behaviour to Your Manager: When, How and With What
This is precisely what I analyse in Your Thoughts Don’t Have to Run Your Life.
There is an asymmetry it is better to know about before starting: a well-prepared approach sometimes changes the situation; a badly prepared one almost always changes it, and in the wrong direction. This is not an injustice, it is a mechanical effect — whoever brings a request stakes their credibility on the solidity of what they bring, and they will not be able to bring it twice.
Hence the purpose of this article: not to encourage or discourage, but to set out what makes such a step admissible.
The threshold: three criteria, and they are cumulative
1. Documented regularity
A single episode, however serious, can always be explained by something else: an exceptional workload, a bad week, a misunderstanding. What cannot be explained away is a series — the same behaviour, with you, over several months, with dates.
This is the most important criterion and the one most often missing. An approach resting on one recent fact plus an undocumented sense of accumulation will be heard as a feeling.
2. A demonstrable effect on the work
A manager can act on what affects the work. They have neither the standing nor the means to act on what belongs to personal affinities.
You therefore need to be able to say: what could not be done, what had to be redone, the time lost, the decision taken on a false basis, the meeting prepared for nothing. Without that, the request is audible but not actionable — and it will be filed as relational, that is, handed back to the two people concerned.
3. A dated direct attempt
"Have you tried talking to him about it?" is the first question anyone will ask, and it is a legitimate question. A vague answer ("I've tried several times") weakens everything else; a precise one ("on 3 April, one to one, I asked that requests come through an identified channel; there were two weeks of change, then a return to the previous pattern") does exactly the opposite.
There is one exception, and it must be stated: when the direct attempt would expose you. It then has to be presented as it is — "I did not open that conversation, for this reason" — rather than inventing an attempt.
The record of facts
It is a short document. One page is enough, and one page is more effective than six.
| Date | What happened | Effect on the work | Record |
|---|---|---|---|
| 12/03, 5.40 p.m. | Correction request sent in the group channel, eight people copied | 2 hrs spent replying publicly, deliverable delayed | the thread |
| 19/03 | Client review held without the date being passed on | Obsolete figures presented, redone the next day | invitation, calendar |
| 02/04 | Direct conversation: request for an identified channel | Two weeks of compliance, then reversion | — |
| 21/04 | Same pattern, same channel | Deliverable redone a third time | the thread |
The four rules that make a record usable
No adjectives. "An aggressive request" can be contested; "a request sent at 5.40 p.m. for 9 a.m. the next day, copied to eight people" cannot. No attributed intent. "He wanted to discredit me in front of the team" is undemonstrable, will be denied, and turns your account into an accusation. "The remark was made in a meeting, in the presence of six people" is a fact. No characterisation borrowed from the law. Do not write the words that belong to the courts. Describe, date, and let everyone do their job — a text that characterises loses its descriptive force and commits whoever wrote it. Chronology rather than anecdote. The order of the table does the work that sentences cannot do: it makes regularity visible. Memory, for its part, keeps only the peaks, and it keeps three of them — which gives, when spoken aloud, the impression of a thin subject.Putting things in series like this can be done by hand, and it is tedious. It is precisely what ScanMyJob produces from professional exchanges: the dates, the channels, the density, the thread exits, the out-of-hours moments. The team conflict case shown as an example shows what a record of thirteen weeks of exchanges between seven people looks like — and what the chronology reveals that the narrative did not.
The meeting
Ask for a slot, stating the subject
A request with no stated subject creates an anxiety that does not help. One sentence is enough: "I'd like thirty minutes on how things are working with X; there's a recurring organisational difficulty and I need a decision."
Note the vocabulary: working, organisational, decision. It announces a professional register and not a settling of scores, which determines how you will be listened to.
Present three facts, not fifteen
This is counter-intuitive and it is decisive. A long account produces two unwanted effects: it makes you look as though you are dwelling on things, and it offers fifteen opportunities to contest a detail — one of the fifteen dates being wrong is enough to make the whole thing doubtful.
Three facts, the most datable and the most demonstrable, are enough to establish regularity. The rest is in the document, which you leave behind.
Word a request, not a grievance
This is the point that separates an approach that gets somewhere from one that bogs down.
A useful request bears on how the work is organised: "I'm asking that correction requests go through the project channel and not the group thread", "I'm asking for the split on this remit to be settled in writing", "I'm asking not to be paired on this type of file any more."
A request that gets nowhere bears on a person or on a sanction: "I'd like someone to have a word with him", "This has to stop", "I want an apology." These requests are not illegitimate; they are simply not within the range of what a manager can grant you, and they put you in the position of asking for a decision that will not be taken in front of you.
Confirm in writing the same day
Three sober lines: what was presented, what was requested, what was answered. Without reproach and without any formal demand.
That written note does two things. It creates a date — the fact that the situation was raised can no longer be contested. And it avoids the divergence of versions that unfailingly appears within forty-eight hours between two people acting in good faith.
The six mistakes that make the step fail
The full chronological narrative. Starting with "it began two years ago" loses attention and shifts the subject on to your endurance. Start with the most recent and clearest fact. The ultimatum. "If nothing changes, I'm leaving" removes your listener's room for manoeuvre and forces them to choose between giving in and losing face. They will almost always choose the second. The request to take sides. Any wording that calls for a judgement about the other person forces the manager into a moral arbitration they will refuse — and they will refuse by sending you both back to sort it out between yourselves, which is exactly what you wanted to avoid. Stacking up witnesses. Arriving with "everybody thinks so" when nobody is prepared to say so weakens your position. A collective approach is possible and effective — provided it really is collective and owned by each person. The informal channel. Raising it in a corridor, at the coffee machine, at the end of another meeting: the format sets the measure of the subject. What is said in passing will be handled in passing. The wrong recipient. If the manager is themselves party to it — because they made the contested call, or because they are involved — going to them is a waste of time and a risk. Another recipient is then needed.When management is not the right person
Three other doors exist, each with its own scope.
And when the situation goes beyond that framework, assessing it is a matter neither for your manager, nor for HR, nor for this article: it is a matter for a lawyer and, ultimately, for a court. What is yours to do is describe dated facts; what is not yours to do is characterise them.
The tests on the platform help you take stock of what the situation is costing you today, and the assistant can help you put a record of facts or a written request into shape before you bring it.Three things this article does not do
It does not tell you to raise it. The three-criterion threshold is there to help you decide, in either direction. It does not describe anyone. A record of facts documents dated behaviours and their effects. It does not say what someone is — and that is precisely what makes it admissible, where a character sketch is inadmissible. It characterises nothing. Neither legally nor clinically. It describes a process and the people available.In brief: A report that fails does not end in the status quo: it ends with a degraded position for whoever brought it. That is why the question is never "should I say something" but "under what conditions can this work". This article first proposes a threshold of three cumulative criteria — documented regularity, demonstrable effect on the work, and the failure of at least one dated direct attempt — without which the step comes too early and rebounds. It then details the preparation of the record of facts: four columns, no adjectives, no attributed intent, no characterisation borrowed from the law — and explains why the chronology is the only genuinely demonstrative element, where an isolated episode can always be explained away. Next comes the conduct of the meeting: announce the subject, present three facts rather than fifteen, word a request that bears on how the work is organised and not on a sanction, and confirm in writing the same day. Finally the article lists the six mistakes that make such a step fail — among them the ultimatum, the full chronological narrative and the implicit request to take sides — and indicates the other people available when management is not the right recipient.

About the author
Gildas Garrec · CBT practitioner
Author of books and articles on applied psychology and relationships.
Join our exchange community on WhatsApp
Join the Community (waiting list)Need personalized support?
Video sessions (€90 / 75 min) or in-office in Nantes.
Book a video session (page in French)Related articles
Resigning in the Middle of Burnout: What That Decision Costs
Leaving ends the exposure, not the exhaustion — and confusing the two is the expensive part. Why a decision to quit taken at the lowest point often backfires, and how to keep it available for later without losing it.
Returning to a Management Role After Burnout
A manager coming back from long-term leave returns to a role whose load is made of availability and arbitration — precisely what lightens least. The four difficulties specific to management, and what gets decided before the return.
Returning to Work After Burnout: A Phased Return in Four Stages
A return to work after long-term leave is decided in the six weeks before the date, not on day one. Four stages, and the three questions to settle before any return date is agreed.
Running Performance Reviews for Your Team: Preparation, Wording, and What Remains Afterwards
From the manager's side: the four biases that distort an appraisal, the three-part feedback formula, how to run a difficult review, and the written record that commits — without ever describing the person.