How do I deal with a boss who never gives feedback but expects perfection

The Monday morning that repeats every week

You send a deck at 4 p.m. Thursday. Nothing comes back. You ship it Friday. On Monday, in a meeting with three other people, your boss says "this isn't what I asked for" — and offers no detail about what was actually wanted. You rework it. The cycle repeats next month with a different deliverable. Over a quarter, you lose somewhere between 15 and 40 hours to rework that a single five-minute conversation would have prevented, and you enter every review cycle without a single documented data point about your own performance.

The specific failure mode here is worth naming precisely, because it is not the same as a mean boss or a micromanager. This is a standards-without-transmission problem. Your boss has a detailed internal model of what "good" looks like. That model is real, it is specific, and it is never externalized. They evaluate against it constantly and communicate about it only when it is violated. Feedback, in their operating system, is an exception handler — it fires on error, never on success, and never in advance.

Three things typically produce this pattern, and they call for different responses:

The overloaded manager. They have 11 direct reports, four projects, and a calendar with no white space. They are not withholding feedback; they have no slack in which to generate it. Diagnostic signal: they respond fast to anything under 30 seconds of effort (a Slack thumbs-up, a one-word reply) and never to anything requiring composition. The fix is to make feedback cheap — reduce the cost of responding to near zero.

The conflict-avoidant manager. They find critical conversations aversive and defer them until the discomfort of the problem exceeds the discomfort of the conversation. That threshold is usually crossed in public, which is why the criticism arrives in a meeting. Diagnostic signal: praise is generic ("great work!"), criticism is delayed and disproportionate. The fix is to create low-stakes, private, scheduled channels where a small correction is easier to voice than a large one.

The "you should just know" manager. Often a strong individual contributor who was promoted without training. Their standard is tacit — they can recognize good work instantly but cannot articulate the rules that produce it. Diagnostic signal: when you ask directly what they want, you get "make it more polished" or "it needs more punch." The fix is to stop asking for abstractions and start showing options, because they can evaluate but not specify.

You will not fix any of these three by asking harder. Asking harder — "can we set up a feedback session?" — puts the composition burden back on the person who has already demonstrated they will not carry it. The whole strategy in this situation inverts that: you carry the burden, and you reduce their required contribution to a yes, a no, or a two-word edit.

How the silence-to-blame loop actually works

The mechanism has a specific shape, and once you can see it, the intervention points are obvious. In the broken loop, work leaves your hands and enters a black box. Nothing returns. You are forced to guess at the standard. You ship on your guess. If the guess matches the hidden model, nothing happens — no signal at all. If it misses, criticism arrives, usually late and usually in a costly setting. Because success generates no signal and only failure generates one, you never accumulate information about what "right" looks like. You only accumulate information about the boundaries of "wrong," one painful sample at a time.

That is the core pathology: the loop only transmits negative signal, so your model of the standard is built entirely out of failures. After four or five of these, most people either freeze (endless polishing, missed deadlines) or disengage (ship anything, stop caring). Both look like performance problems to an outside observer, which is how a communication failure gets recorded as a competence failure in your file.

The intervention breaks the loop at one point: you move the decision before the work instead of after it. Instead of shipping a finished artifact and waiting for a verdict, you ship a cheap statement of intent and let silence carry meaning.

The mechanics of the written pre-commitment:

1. Before starting anything above roughly four hours of effort, send a 60–120 word note. Structure: what you're building, the three assumptions you're making, the date you'll deliver, and one explicit line — "If I don't hear otherwise by [date], I'll proceed on these assumptions."
2. Make the assumptions falsifiable and specific. "Assuming we're targeting enterprise accounts only, using Q3 pipeline data, and that a 12-slide format is right" is checkable in eight seconds. "Assuming the usual scope" is not.
3. Give a real deadline, 24–48 hours out. Not "let me know when you can." A named date is what converts non-response into a decision.
4. Send it in the channel they actually read. If they answer Slack and ignore email, use Slack. Match their medium, not your preference.
5. Keep a copy. Email to yourself, a doc, a Notion page — anywhere outside a system your employer can wipe on your last day.

The reason this works on all three manager types is that it costs them almost nothing. The overloaded manager can reply with a single word. The conflict-avoidant manager gets to correct an assumption rather than criticize a person — a much lower-friction act. The "you should just know" manager gets something concrete to react to instead of an abstraction to articulate.

The second structural piece is the weekly written summary. Five to eight bullets, sent the same day and time every week: what shipped, what's in flight, what's blocked, and one line of "here's what I'm assuming about priority for next week." This is a fifteen-minute habit that produces the single most valuable artifact you can own in this situation — a timestamped, contemporaneous record of your work and your stated assumptions, created before any dispute existed. If a performance conversation ever goes sideways, contemporaneous documentation is worth vastly more than reconstructed memory.

What the numbers actually look like

Precise public statistics on this exact pattern are thin, and you should be suspicious of anyone quoting a hard percentage. But the operational numbers — the ones you control — are worth being concrete about, because vagueness is what makes this problem feel unsolvable.

Time cost of the current state. Track it for two weeks. Log every hour spent on rework that a pre-brief would have prevented, plus every hour spent in anxious polishing driven by uncertainty rather than by the work itself. Most people running this exercise land somewhere between 3 and 8 hours per week. At the low end of that range, over a 46-week working year, that is roughly 140 hours — about three and a half working weeks annually spent on a communication gap. That number is your business case, and it is the number to bring to any escalation conversation.

Time cost of the fix. The pre-commitment note takes 6–10 minutes to write. If you send two or three a week, that is 20–30 minutes. The weekly summary takes 15 minutes if you keep a running scratch file during the week and 45 if you reconstruct it Friday afternoon — keep the scratch file. Total overhead: roughly 45 minutes per week, or about 1.5% of a 40-hour week. If it prevents even one hour of rework weekly, the return is better than 100%.

How long to give it before escalating. Run the system cleanly for a full quarter — 12 to 13 weeks — before concluding it hasn't worked. Two weeks is not a test; managers are erratic and any two-week window is noise. At the 6-week mark, do a mid-point check: are you getting *any* response rate above zero on the pre-commitments? Even a 20–30% response rate is a working system, because the other 70% converts to documented approval, which was the point.

Response-rate benchmarks worth tracking. Keep a simple tally. Column one: pre-commitments sent. Column two: replies received. Column three: instances where late criticism arrived on something you had pre-committed in writing. That third column is the one that matters. If it drops toward zero over a quarter, the system is working even if the reply rate stays low. If late criticism keeps landing on things you explicitly flagged in advance, you have a different and more serious problem — one that is about the manager's fairness rather than their bandwidth, and that is the trigger for escalation.

Meeting cadence. If you have a recurring 1:1, protect it aggressively and use the last 10 minutes for a standing question. If you don't have one, ask for 20 minutes biweekly — a smaller ask that is harder to refuse than 30 minutes weekly. The lowest-friction question to ask at the end of every one, in these exact words or close to them: *"What's one thing I should do differently next time?"* It is bounded to one item, it presumes there is something (which makes silence awkward rather than comfortable), and it is future-facing rather than an invitation to critique the past. If the answer is "nothing," follow with *"If you had to pick one, even a small one?"* Ask it every single time. The fourth or fifth ask usually produces something, because the pattern of the question makes non-answering the effortful choice.

The "show three options" technique, with a time budget. For the "you should just know" manager, stop asking what they want and show them cheap versions. Three rough options, 20 minutes of work each — not three polished deliverables. A rough option is a bullet outline, a napkin wireframe, a two-paragraph sketch. Present them as A, B, and C and ask which direction to develop. People who cannot specify a standard can almost always pick one from a set. Budget an hour total; if you're spending four hours making options beautiful, you have reintroduced the exact rework you were trying to avoid.

A concrete week. Monday: send the weekly summary with next-week assumptions (15 min). Tuesday/Wednesday: send pre-commitments for any new work above four hours (10 min each). Thursday: in the 1:1, ask the one-thing question (30 seconds). Friday: update your running scratch file and your tally (10 min). Total: under an hour, every week, permanently.

What you trade away, and what else you could do

None of this is free, and it is worth being honest about the costs before you commit a quarter to it.

The main trade-off is that you look process-heavy. Some managers read a stream of written pre-commitments as bureaucratic, defensive, or as a sign you can't operate independently. Mitigate it with tone and length. Keep notes under 120 words. Write them as a heads-up, not as a request for approval: "Heads up — starting the pipeline analysis today, scoping it to enterprise only and using Q3 data, delivering Thursday. Shout if that's wrong." That reads as autonomy. "Can you please confirm the scope before I begin?" reads as dependency. Same information, opposite signal.

The second cost is emotional. Sending careful notes into silence, week after week, is genuinely draining. You are doing the relational labor for two people and receiving nothing back. Budget for that. The system protects your record and reduces rework; it does not make you feel valued, and expecting it to will leave you disappointed. Get the recognition need met elsewhere — peers, a mentor, external community — and treat the manager relationship as a logistics problem rather than a source of validation.

The third cost is opportunity. A quarter spent optimizing around an absent manager is a quarter not spent somewhere with a functioning one. That is a real cost, and it should be weighed honestly rather than deferred indefinitely.

Here is the decision structure across the realistic alternatives:

Alternative one: lateral and downward feedback. If your manager will not evaluate your work, get evaluation from people who will. Ask two or three peers whose judgment you respect for a 20-minute review of something specific — not "how am I doing," but "I'm about to send this; what would you change?" This is a genuine substitute for the technical part of feedback. It does not substitute for the political part, because peers cannot tell you how your work is perceived by the person who writes your review.

Alternative two: the skip-level. Riskier, and the sequencing matters. Do not use a skip-level to complain about your manager — that is read as an end-run and usually rebounds. Use it to ask about standards: "I want to make sure I'm calibrated on what good looks like for this team. What separates strong work from adequate work here?" That is a legitimate question from a motivated employee, it produces genuinely useful information, and it puts a data point about your engagement in front of someone senior. Do it once or twice a year, not monthly.

Alternative three: formal escalation. Reserve this for the case where documented pre-commitments are being overridden by retroactive criticism — that is, where the fairness problem is established in writing. Bring the tally, the dated notes, and the hours-of-rework number. Frame it as a workflow problem you're trying to solve, not as a complaint about a person. HR responds far better to "here's a documented pattern costing about 140 hours a year, here's what I've tried" than to "my boss never gives me feedback."

Alternative four: leave. If a full quarter of a clean system produces no change, and a skip-level produces no change, the honest read is that this environment will not give you what you need. That is information, not failure. The written record you built is portable — it is the raw material for your resume bullets and your interview stories, which is a second, quieter return on the whole exercise.

The mistakes that make this worse

Asking "do you have any feedback for me?" This is the single most common move and the least effective. It is unbounded, it requires the other person to generate content from nothing, and it is trivially answered with "you're doing great." Replace it with a bounded, specific question: "On the Q3 deck specifically — was the length right, or should I have cut it to ten slides?" One artifact, one dimension, a binary-ish answer. Bounded questions get answered; open ones get deflected.

Escalating too early. Going to HR or a skip-level in week three, before you have documentation, puts you in a he-said-she-said with no evidence. You will be characterized as difficult and the underlying problem will persist. Build the record first. The record is the leverage; without it you have only an assertion.

Letting silence be ambiguous. If your notes do not contain an explicit "I'll proceed unless I hear otherwise by [date]," silence means nothing and protects you not at all. That single sentence is what converts non-response into a decision. Do not soften it into "let me know if you have thoughts" — that has no deadline and no default, so it carries no weight later.

Chasing perfection to preempt criticism. The most damaging response to this dynamic is to try to be so flawless that criticism becomes impossible. It never works — the standard is hidden, so you cannot hit it reliably — and it costs enormous time while producing visible anxiety. Set a defensible quality bar, hit it, ship, and let the pre-commitment note carry the risk. If criticism comes anyway, you have the note.

Venting to colleagues instead of documenting. Complaining feels like action and is not. It also travels; assume anything you say about your manager reaches them eventually. Channel the energy into the weekly summary, which is the same effort and produces an asset instead of a liability.

Going silent in return. When feedback stops, a natural response is to stop communicating too — fewer updates, less visibility, quiet work. This is exactly backwards. An absent manager fills the information vacuum with assumptions, and those assumptions are rarely generous. Increase your outbound communication as their inbound decreases. Being the person whose status is always legible is a real form of protection.

Making it a personality conflict. The moment this becomes "my boss doesn't like me," you lose the ability to solve it as a process problem, which is the only frame where you have leverage. It may well be true that they don't like you. It is still more useful to act as if the problem is bandwidth and transmission, because that framing produces actions you can take.

Waiting for the annual review to raise it. By then the record is written. Everything here works because it is contemporaneous. A concern raised in month two is a workflow conversation; the same concern raised at review time is a defense against a rating, and it will be heard that way.

Sending the system inconsistently. Three pre-commitments in week one, none in weeks two and three, one in week four. Irregular signals get filtered as noise. Same day, same time, same format — the predictability is what makes people eventually start reading them, and it is what makes the record credible if you ever need it.

Related questions

How do I ask for feedback without sounding insecure?

Anchor the question to a specific artifact and dimension rather than to yourself. "Was the deck too long?" reads as professional calibration. "How am I doing?" reads as reassurance-seeking. Bounded, work-focused questions signal confidence; open, person-focused ones signal anxiety.

What if my boss only gives feedback in front of other people?

Address it privately and once, framed as effectiveness rather than feelings: "I take the input better one-on-one — can we do course corrections in our 1:1?" Then keep pre-committing in writing, which shifts corrections earlier and out of the meeting entirely.

Should I go to HR about a manager who never gives feedback?

Not on its own — no feedback is rarely actionable for HR. Escalate only when you can show documented pre-commitments being overridden by retroactive criticism, with dates and an hours-of-rework figure. That is a pattern; the absence of praise is not.

How long should I stay in a role with no feedback?

Give a documented system one full quarter, then a skip-level conversation, then 60 days. If nothing shifts across roughly five to six months of deliberate effort, the environment is unlikely to change and staying costs you calibration you can't get anywhere else.

Does this change if my boss is remote?

It gets worse and the fix gets more important. Remote removes the incidental signals — hallway comments, expressions in meetings — that partially substitute for real feedback. Written pre-commitments and weekly summaries become the entire channel rather than a supplement to it.

FAQ

Isn't sending written notes going to annoy my manager?

Only if they're long or read as requests for permission. Keep them under 120 words, write them as heads-up notices rather than approval requests, and require nothing from the reader in the default case. A manager who is annoyed by a 90-word note stating your plan and your delivery date is telling you something important about the relationship, and that information is itself useful.

What if my boss explicitly tells me to stop sending updates?

Ask what they'd prefer instead and take the answer seriously — some managers genuinely want less inbound. Then reduce to the minimum viable version: one weekly summary, no pre-commitments except on work above roughly a week of effort. Keep your own private dated log regardless. The record is for you, and it does not require their participation.

Does the "silence means approval" line actually protect me?

It is not a legal shield, and it will not stop a determined manager from criticizing you anyway. What it does is change the conversation. "I flagged these three assumptions on the 14th and didn't hear back" moves the discussion from your judgment to a process gap, which is a much better position. It also matters cumulatively — a quarter of dated notes is far more persuasive than one.

How do I handle a boss who expects perfection but won't define it?

Stop asking for the definition and start proposing one. Write down what you believe the standard is — specific, checkable criteria — and send it as your working assumption. People who cannot articulate a standard can almost always correct a wrong one. You are converting an authoring task, which they will not do, into an editing task, which they often will.

What if I get criticized for something I explicitly pre-committed on?

Once is worth absorbing calmly — reference the note briefly and move on. A repeated pattern is the real signal, and it changes the diagnosis from bandwidth to fairness. That is the point at which the documentation stops being a workflow aid and becomes the basis for a skip-level or HR conversation. Track those instances in a separate column; the count is the argument.

Can I fix this without my manager changing at all?

Largely, yes, and that is the design intent. Every element here works whether or not they participate: pre-commitments reduce rework because they force you to make assumptions explicit to yourself; weekly summaries build a record regardless of readership; lateral feedback supplies genuine calibration. Their engagement improves the outcome but is not required for the system to pay for itself.

Sources

• https://hbr.org/2019/03/the-feedback-fallacy
• https://hbr.org/2016/01/how-to-get-the-feedback-you-need
• https://www.gallup.com/workplace/236570/employee-feedback-not-enough.aspx
• https://www.mindtools.com/a0i9ero/how-to-get-feedback-from-your-boss
• https://www.shrm.org/topics-tools/news/employee-relations/performance-management-documentation
• https://hbr.org/2022/09/what-good-feedback-really-looks-like
• https://www.mckinsey.com/capabilities/people-and-organizational-performance/our-insights/the-boss-factor-making-the-world-a-better-place-through-workplace-relationships
• https://www.apa.org/topics/healthy-workplaces/manager-communication

Related on PULSE

• [How do I push back on an unrealistic deadline without looking difficult?](/knowledge)
• [What should I document before a performance review goes badly?](/knowledge)
• [How do I run a 1:1 with a manager who cancels half of them?](/knowledge)
• [How do I get promoted when my manager doesn't advocate for me?](/knowledge)
• [When is it worth escalating to a skip-level manager?](/knowledge)