A Bad Performance Review: The Next 48 Hours

There is a difference in kind between the two hours that follow a difficult review and the two days. In those two hours you have access neither to a faithful memory nor to your judgement: the physiological state produced by negative appraisal is the same as the one produced by a threat, and it literally narrows the range of what you can envisage.

Which is why the first rule is about what not to do.

Day 0, that evening: send nothing

No message, no email, no reply to the written record. What gets written on the first evening is always either too long, or too hard, or too accommodating — and it stays. Don't recount it either. The urge to give your version to two or three colleagues is very strong, and it costs you: the version circulates, it comes back distorted, and it installs a reading of you in the team that will precede you into every room. The only legitimate audience on the first evening is someone outside the organisation. One thing only is worth doing, and it is useful: write for yourself. Thirty minutes, with no structure, everything that was said — the exact sentences as far as possible, the order, the examples cited, the moments when the tone changed. That document is not addressed to anyone. It serves one precise purpose: giving you, tomorrow, what was actually said, before the night has rewritten it.

Because that is what happens every time. Within twenty-four hours, negative feedback transforms itself in the memory: the qualifications drop away, the examples merge into a global judgement, and a sentence along the lines of "that particular point didn't work" becomes "he told me I wasn't up to it". The evening's record is the only protection against that drift, and it often changes the reading completely by the next day.

Day 1: sorting

Take your record and put each item into one of three columns. It is a mechanical exercise, and it is the heart of the work.

Column 1 — Well-founded

What is accurate, however unpleasant. There is almost always at least one, and acknowledging it to yourself is not a capitulation: it is what makes the rest credible.

What you do with it: you register it, you identify what concretely changes, and you say so in the written reply. One point acknowledged, precisely, with an action attached, is worth more than ten points defended.

Column 2 — Imprecise

What is worded too generally to be checked or corrected: "lack of proactivity", "your manner needs work", "not present enough on cross-functional topics".

What you do with it: you ask for detail, in writing, without aggression. "On the point about presence on cross-functional topics, could you point me to two or three specific situations? I need to know where to put the effort."

That request does two things. It turns a judgement into a piece of work — which is in your interest. And it produces information: either examples arrive, and you know what to do; or they don't, and that says something about how solid the assessment is.

It is the most important of the three columns, because an appraisal that rests on no stated criterion is neither contestable nor correctable. The research on procedural justice identifies precisely this point as decisive: it isn't the severity of an appraisal that damages a working relationship, it's the fact that it is ungrounded and not open to discussion.

Column 3 — Factually disputable

What is factually inaccurate: a date, a deliverable credited to someone else, a delay whose cause is documented elsewhere.

What you do with it: you reply, once, with the item. "File X was sent on the 11th, here is the email." No commentary, no indignation, no "I'd like to point out that". The fact is enough and it is stronger on its own. How many: two or three at most. Contesting a written record point by point produces the opposite of the intended effect — it reads as a refusal to hear, and it earns the person who writes it a reputation that outlasts the appraisal being contested.

Day 1: the sorting that decides everything — the content or the manner?

A second distinction, less well known and very useful. What hurt you falls into one of two registers, and they are not handled together.

The content: the appraisal itself, the rating, the points raised, the consequence for pay or progression. The manner: the fact that you couldn't reply, that it was said in front of someone, that the tone was contemptuous, that the conclusion had plainly been written before the conversation, that criteria had never been announced.

Research distinguishes these two registers — procedural justice for the process, interactional justice for how a person is treated — and shows that they have different effects. The second produces more intense and more lasting reactions than the first.

Why this is useful in practice: many people contest the content when what got to them was the manner. The challenge fails — because the content may well have been defensible — and what actually hurt is never addressed. Separating the two lets you answer the content with facts, and raise the manner as a distinct subject, at another moment, in the words that suit it.

Day 2: the written reply

This is the document that will remain, and that will be read by people who weren't in the room. It runs to ten or fifteen lines.

Its structure:

One sentence acknowledging receipt, neutral. "I have received the record of the review of the 14th."

The point acknowledged, with what you are doing about it. "The point on response times is well-founded. I am putting a weekly follow-up in place from next week."

The requests for detail on the imprecise points, one or two.

The factual corrections, two or three at most, with the item attached.

A request for a follow-up point at a set date: "I suggest we review these subjects again at the end of March."

What it doesn't contain: any explanation of your state, any reproach about the manner — that gets said, not written — any ultimatum, and no copying in of third parties who aren't concerned. That last point matters: widening the circulation of a reply to an appraisal is perceived as an escalation, whatever the content. On formalities. Many people wonder what signing or not signing a written record means, about deadlines, about what can be added. Those questions depend on a text, a collective agreement or a custom, and are put to your HR department, to an employee representative, or to the labour authority in your country. No article can answer them in their place — and a decision taken on the strength of something found online is what produces the most damage in these situations.

The mechanism to watch: attribution

What decides the effect of a negative review three months later isn't its content: it is the explanation you give of it. Bernard Weiner's work established that three properties of that explanation change everything.

Internal or external. "It's me" or "it's the situation". Stable or unstable. "That's how it is" or "that was this year". Global or specific. "I'm worthless" or "that particular point didn't work".

The combination that does the most damage is internal, stable and global: I'm bad, I always have been, it won't change. It appears spontaneously in the first forty-eight hours, it looks clear-sighted, and it is almost always false — because no review is about what someone is.

The three-column sort proposed above doesn't only have a practical function: it is the best antidote to that drift. It mechanically brings a global judgement back to three or four specific, dated points, some of which turn out to be wrong anyway.

The opposite marker to watch for — the entirely external attribution, "they have a problem with me, none of it is true" — has the advantage of protecting you in the moment and the drawback of making any correction impossible. Both extremes cost, and the sorting avoids both.

What helps in the days that follow

Recovering the actual year. A negative appraisal crushes the memory of the preceding twelve months: after a bad review, you can no longer recall anything that worked. Your work correspondence, however, has kept it — the requests handled, the feedback received, the periods of load, the warnings raised. Re-reading them isn't consolation: it is a correction of measurement, and it is also the material for the next review. That is what ScanMyJob puts back into sequence from the exchanges themselves, without adding any conclusion. Don't decide anything for two weeks. The rate of resignation decisions taken within ten days of a negative review is high, and those decisions are taken in the least suitable state there is. Nothing gets worse by waiting a fortnight; a great deal becomes clearer. Watch your sleep. If the feedback is still going round at night after a week, if apprehension about going to work settles in, this is no longer a question about a review. A doctor is the right person — an occupational health service where your country provides one, otherwise your own doctor — and medical confidentiality applies. The tests on the platform let you take stock of your relationship with appraisal and criticism — two equally competent people absorb the same feedback in completely different ways, and that difference explains a good part of what is happening here. The assistant can help you shape the written reply.

Three things this article does not do

It does not say how to challenge formally. The routes of appeal, the deadlines and the formalities depend on a precise framework: those are questions for your HR department, an employee representative, the labour authority in your country, or a lawyer. It characterises nothing. A difficult review is not a legal category, and judging what would go beyond that framework belongs neither to this article nor to you: it belongs to a court. It does not describe anyone. Neither you nor the person who conducted the review. What gets sorted is dated statements and verifiable facts.