How to Manage Up as an Investment Banking Analyst

Introduction

Nobody runs a training session on how to talk to the people who staff you. You get taught Excel shortcuts, formatting standards, and the firm's compliance rules, and then you are dropped onto a deal team where three people at three different levels all want something from you at once, and none of them will explain how they want to be handled. Working that out is called managing up, and it is the widest gap between analysts who are technically competent and analysts whom seniors compete to staff.

Managing up is not politics and it is not flattery. It is the operational skill of being easy to rely on: taking instructions cleanly, reporting status before anyone has to ask, surfacing problems while they are still cheap to fix, and giving each level above you the particular kind of certainty that level needs. It is learnable, it is mostly language, and almost none of it is intuitive when you are twenty-two and terrified of looking incompetent.

This post is about the working relationship specifically, not about overall performance. If you want the broader picture of how banks rank a class and what drives your bonus, our breakdown of what actually makes a top-bucket analyst covers that ground. What follows is narrower and more practical: what each senior wants, the exact scripts that work when you need to flag, trade, or apologize, and what to do when the person above you is the problem.

The single most common junior mistake is treating the associate, the VP, and the MD as one undifferentiated mass called "seniors." They are not. They sit at different distances from the client, they get judged on different things, and each one is afraid of a different failure. Once you understand what each of them is protecting, most of the guesswork about how much detail to send and how fast to reply disappears.

Managing Up

Managing up is the deliberate work of building a productive relationship with the people you report to: learning how each of them prefers to receive information, setting expectations in both directions, and negotiating priorities before they collide. It has been treated as a core professional skill since John Gabarro and John Kotter's Managing Your Boss first appeared in Harvard Business Review in 1980, and it is worth more in banking than almost any other entry-level job because the feedback loops are so fast.

The table below is the compressed version. It is worth reading before every staffing, because the same deliverable gets communicated three different ways depending on who is receiving it.

The Associate Wants Accuracy and No Surprises

Your associate is the person who checks your work before anyone senior sees it, which means every error you make becomes their problem first. When a number in your comps page is wrong, the VP does not call you. They call the associate. That structural fact explains almost everything about associate behavior: the nitpicking, the re-checking, the irritation when you go quiet.

What earns you an associate's trust is boring and specific. Numbers that tie to a source you can name. Footnotes that match the page. Version files that follow the naming convention rather than ending in "final_v2_REAL." A one-line note when you send work back saying what you checked and what you did not. Associates are not looking for brilliance from a first-year, they are looking for the absence of surprises, and every surprise costs them a night.

The VP Wants the Deck to Hold Together

The VP owns the story. They are the person who has to stand in front of a client and walk through a book that makes a coherent argument, and their fear is not a wrong decimal, it is page fourteen contradicting page six. A VP will forgive a formatting slip that an associate would flag; they will not forgive a valuation summary whose implied range does not match the football field two pages earlier.

Communicating with a VP means leading with the conclusion and the exceptions. "Comps are updated, three names moved on earnings, the median multiple is down about half a turn, which pulls the low end of our range down and I have already reflected that on the summary page." That is a complete VP update. Sending them the file with "done" attached forces them to do the reading you were supposed to do, which is exactly the work they staffed you to absorb.

The MD Wants to Never Be Embarrassed

The MD is closest to the client and furthest from the file. They will not open your model. What they need is to walk into a room, get asked a question, and have an answer that is right. Their nightmare is a client saying something they did not know, or a number in the book they cannot defend.

That means when an MD asks you something, the answer is one line first and detail only if requested. "Revenue growth was eleven percent last year" beats a paragraph about the adjustment you made for the divested segment, even when the adjustment is the interesting part. Mention it second. If an MD wants the mechanics, they will ask, and the ability to produce the mechanics on demand is what makes you the analyst they request by name. The wider view of who does what at each level is laid out in our guide to the investment banking hierarchy from analyst to MD.

Most rebuilds do not come from bad analysis. They come from a request that was heard incompletely and executed confidently. A junior who takes ninety seconds to pin down what is actually being asked will out-deliver a faster junior who guesses, every single time, and the ninety seconds is almost entirely language.

Repeat the Request Back Before You Start

Before you touch anything, say the task back in one sentence in your own words. Not "got it." Something like: "So you want a three-page summary of the four precedent transactions with EV/EBITDA and the premium paid, formatted like the pages in the last book, for tomorrow morning." Roughly one time in four, the senior will correct a piece of that sentence, and each correction you catch here is an hour you do not lose tonight.

This feels awkward for about a week and then it becomes invisible. Seniors do not read it as slow. They read it as a person who is going to come back with the right thing.

The Two Questions That Prevent a Rebuild

There are two questions that account for most avoidable rework, and you should ask both on any task longer than an hour:

A third question is worth asking when the request is analytical rather than production work: what would make this analysis wrong? Seniors usually know the trap already, and saying it out loud saves you from walking into it.

The Stated Deadline and the Real Deadline

"End of day" is not a deadline. It is a placeholder that could mean 6 p.m., could mean before the associate logs off at 1 a.m., and could mean before a 7 a.m. flight. The real deadline is set by an external event: a client call, a board meeting, a printer cutoff, a filing. Find the event and you find the deadline.

The question is easy to ask without sounding like you are negotiating: "When do you need to have reviewed it, and when does it go out?" Those are two different times, and the gap between them is the buffer you are actually working against. A deliverable that lands at 11 p.m. for a VP who wanted to turn it before midnight is fine. The same deliverable at 11 p.m. when the VP planned to review at 9 is a failure, even though nobody said 9.

Pencils Down

A pencils-down period is a protected window in which juniors are not expected to work or respond. JPMorgan's runs from 6 p.m. Friday to noon Saturday and sits alongside a cap of 80 hours a week and a guaranteed full weekend off each quarter; Goldman Sachs runs a wider Saturday rule covering 9 p.m. Friday to 9 a.m. Sunday. The 80-hour cap carries an explicit live-deal exemption and the windows bend the same way in practice, so the practical skill is knowing whether a request lands inside a protected window and, if it does, who authorized the exception.

Of everything in this post, the status update is the highest return per second spent. Silence is the fastest way to damage senior trust, faster than a wrong number, because a wrong number is a mistake and silence is an unknown. An associate who has not heard from you cannot tell whether you are ninety percent done or have not started, so they assume the worse case and either chase you or, much worse, quietly start rebuilding your work themselves.

The Three Moments That Always Need an Update

You do not need to narrate your night. You need to send something at three specific moments:

That is three messages across an eight-hour task, and it is the difference between an associate who leaves you alone and an associate who checks in every forty minutes because they cannot afford not to.

What a Useful Update Actually Contains

A useful update has three parts: where you are, what has changed, and when it lands. "Still working on it" has none of them and is worse than nothing because it consumes attention without transferring information. The update that builds trust is specific enough that the senior can make a decision without replying to you.

If something has slipped, the update carries the new time, not an apology. "This is going to be closer to 11 than 9, the data pull is slower than expected, still on track for your review before you turn it" gives the VP a chance to re-plan. Discovering the slip at 9:05 does not.

Junior bankers systematically under-flag, because every flag feels like an admission that you cannot handle it. The result is that problems surface at the worst possible moment, when they are expensive and public rather than cheap and private. Learning to raise things early, in language that reads as ownership rather than resistance, is most of what separates an analyst who gets trusted with hard work from one who gets given safe work.

Flag Early, and Always Flag With a Proposal

The rule is simple: the moment you know something will not work, say so, and bring an option with you. A flag without a proposal is a complaint. A flag with a proposal is management.

Compare "I do not think I can get this done by tomorrow" with "To hit tomorrow morning I can either give you the four core names fully built, or all nine names with multiples only. Which is more useful?" The second version transfers a decision, not a burden. It also demonstrates you understand what the work is for, which is the thing seniors are quietly testing.

Language That Works and Language That Backfires

The phrasing genuinely matters, and the pattern is consistent. Language that works is specific, offers a path, and keeps you on the same side of the problem as the senior:

The last one is the real killer. Silent overcommitment is treated as a character issue, not a capacity issue, because the senior planned their own night around your yes.

Trade Instead of Refusing

You almost never have the standing to decline work outright as a first-year, and you almost always have the standing to trade. Trading means putting your existing commitments on the table and letting the person with the fuller picture prioritize. "Here is what I am holding, here is what this would cost, you choose" is not pushing back. It is handing over a scheduling problem you are not senior enough to solve alone.

Managing seniors starts in the interview, not on day one: Interviewers test judgment under pressure with behavioral questions about conflict, deadlines, and difficult teams, so start practicing interview questions for free and see which of your answers still sound rehearsed.

Sooner or later a VP on one deal and an associate on another will both need your night, and neither of them knows the other asked. This is the most common conflict a first-year faces, and it is also the one juniors handle worst, because the instinct is to quietly try to do both and fail at both.

The move is to make the conflict visible rather than to arbitrate it. You are not senior enough to rank a VP's pitch against an associate's model, and attempting it means one of them finds out later that you deprioritized them without telling them, which is far worse than the conflict itself. Reply to each with the same information: what else you are holding, what the timings are, and a specific question about what should give.

Where a genuine tie-break is needed, or where the two people are unlikely to talk to each other, route it through the person whose job is exactly this. Escalating to the staffer is not tattling, and it should be phrased as a routing question rather than a grievance: "Both of these landed for tomorrow morning. Can you help me sequence them?"

Staffer

The staffer is the senior associate or VP who allocates juniors to deals within a group and holds the full picture of who is working on what. In a conflict between two seniors who both want the same night, the staffer is the neutral party with the authority to decide, which is why routing a clash to them is a scheduling step rather than a complaint. Our post on how staffing works in investment banking covers how those assignments are actually made.

When the Staffer Has to Get Involved

Escalate when the two requests are genuinely simultaneous and both are hard deadlines, when you have already told both parties and nothing moved, or when the pattern repeats across a week rather than happening once. Do not escalate the first minor overlap you hit, and do not escalate as a way of avoiding a conversation you should have had directly. The staffer's patience is a resource, and juniors who spend it on ordinary Tuesday-night collisions have none left when a real problem arrives.

Every deck you touch will come back covered in marks, and how you respond to that is watched more closely than you think. The default junior reaction, explaining why you did it the way you did, reads as arguing even when it is genuinely informative, and it is almost always the wrong opening move.

Turns

A turn is one complete round of edits on a document. A senior marks up a draft, the junior makes every change, and a clean version goes back for the next review. Deal teams count work in turns rather than hours: saying a book "has three more turns in it" means three more full review cycles before it reaches the client, which is why an analyst who clears a turn cleanly the first time saves the whole team a night.

The mental reframe that helps most is that a turn is not a verdict on you. It is the production process working. A book that goes to a client has typically been through five to ten turns, and the marks on your page are the same marks that were on the VP's page when they were an analyst. Research on how high performers handle critique consistently finds that the useful move is separating the information in the feedback from the emotional reaction to receiving it, a point Harvard Business Review has made in its work on how to manage feedback. In banking the practical version is: make the changes first, and ask questions afterwards.

The Three Legitimate Responses to a Comment

When you get a marked-up page back, every comment falls into one of three buckets, and each has a correct response:

The third case is where reputations are built, because a junior who can raise a substantive objection calmly and then execute without sulking is genuinely rare. What kills you is arguing twice, or making the change while visibly resenting it.

Ask for the Reasoning, Just Not in the Moment

Understanding why a senior wanted something is how you stop needing the comment next time, but 11 p.m. during a turn is not when to ask. Bank the question and raise it the next day when the pressure is off: "You moved the sensitivity table to the appendix on Tuesday. Is that a general preference or specific to that client?" Asked calmly and later, that question makes you look like someone building a system. Asked mid-turn, the same question sounds like resistance. The related behavioral question comes up in recruiting too, and our walkthrough of how to answer tell me about a conflict with a coworker covers the same instinct in interview form.

Every senior banker made errors as a junior and they expect you to make them. What they do not expect, and do not forgive, is finding out that you knew about a problem and stayed quiet. A mistake is an operational event with a cost measured in hours. A hidden mistake is a trust event, and the cost is measured in how much independent work you get for the rest of the year.

The Self-Report Script

The self-report script is short and should be delivered the moment you find the issue, regardless of the hour: what is wrong, what it affects, what you have already done, and what you need. "The EBITDA in the comps page pulled the reported figure instead of the adjusted one for two names. It moves the median about a tenth of a turn. I have rebuilt the page and it is attached; the summary on page four needs the same fix, which I can do now. It went out to the client this morning, so tell me how you want to handle that." No preamble, no lengthy apology, no explanation of how it happened until asked.

Your review is not written by one person. Feedback gets collected from every senior you worked with and then argued out in a room, so an analyst adored by a single VP and unknown to everyone else is in a weaker position than one whom four people describe as reliable. Managing up well with one associate is a start; the goal is a group of seniors who all reach for the same word about you.

That breadth comes from practical habits rather than visibility-seeking. Take staffings across a range of associates and VPs rather than becoming one person's exclusive resource. Hand off cleanly when you roll off a deal, with a note covering file locations, open items, and what the next person needs to know, because the successor's opinion of you is formed entirely by that handoff. Ask for a short piece of feedback mid-deal instead of waiting for the formal cycle, which is both useful and, quietly, a signal that you are trying. And be decent to the analyst next to you, since peers become associates and the person you helped at 1 a.m. is in the room two years later. That progression is worth understanding early, and our post on the analyst-to-associate promotion path explains how those relationships compound.

A 160-page reference you can actually work through: Our PDF walks through the technical frameworks, valuation mechanics, and behavioral answers that carry from recruiting into the seat, so access the IB Interview Guide and use it as your baseline before day one.

Not every problem is yours to fix by communicating better. Some associates give instructions that change every two hours, hold work for six hours before turning it, or hand you a task at 7 p.m. that they have been sitting on since noon. Bad management exists, and pretending otherwise leads juniors to absorb blame for structural problems.

Diagnose Before You React

Start by working out which kind of bad you are dealing with, because the responses differ completely. A disorganized associate needs structure from you: written confirmations of every verbal instruction, a running list of open items you send at the start of each day, and explicit "you asked for X on Monday, is that still what you want?" check-ins that create a paper trail without being aggressive. A genuinely hostile associate is a different situation and does not respond to better process at all.

Documentation is the quiet defense in both cases. Confirm instructions by message rather than only in conversation, keep your own log of when work was requested and delivered, and note who changed what. You are not building a case. You are removing the ambiguity that makes it easy for a bad manager's disorganization to become your performance problem. Hours are more visible than they used to be too: JPMorgan is piloting a program that compares self-reported timesheets against system activity and sends juniors weekly reports on their actual hours, which makes chronic overload harder to wave away when you do raise it.

What Is Actually Worth Escalating

Ordinary friction is not an escalation. Being given a hard task, getting blunt comments, or working a bad weekend on a live deal is the job, and framing any of it as a complaint marks you as someone who does not understand the environment. What is worth escalating, through the staffer, an HR contact, or a senior who has taken an interest in you, is a pattern: repeated last-minute work that was sat on, hours that stay extreme when no deal is live, credit for your work taken publicly, or anything abusive.

When you do raise it, bring dates and deliverables rather than adjectives, and lead with the effect on the work. "Over the last three weeks I have received four Friday-evening requests on work that was ready Wednesday, and it has cost the Riverside timeline twice" lands. "The associate is disorganized and it is unfair" does not, however true it is. Meanwhile, protect what you can control: the deals you deliver on, the seniors who can speak to your work, and the record. Most bad staffings end within a quarter, and the analysts who survive them well are the ones who did not let a single difficult relationship shrink their reputation to one data point. If you are still pre-start and want a picture of the rhythm you are stepping into, our look at a day in the life of an analyst sets the baseline.

Managing up sounds like soft skills and it is not. It is a set of concrete, repeatable behaviors: a sentence that confirms the ask, a question that finds the real deadline, a message at the midpoint, a trade instead of a refusal, a comment taken without argument, an error surfaced before anyone finds it. Each one takes seconds. Together they decide whether the people above you have to think about you, which is the only variable that really matters in a junior banking seat.

The analysts who figure this out early are not more talented than their class. They are the ones who realized that technical work is the entry ticket and the relationship is the job, and who treated managing their seniors as a craft to practice rather than a personality trait they either had or did not. Start with the request script and the status update. Those two alone will change how you are staffed within a month, and everything else in this post gets easier once the people above you have stopped worrying about whether the work is coming.