Most managers are trained to look downward, at the team they direct, the work they assign, the people they appraise. But almost nothing important gets done through that line alone. The budget sits with a peer in finance, the green light sits with your boss, and the dependency you need sits in a department you have no authority over at all. Managing up, down and across is simply the recognition that leadership runs in three directions, and that the two you don't command are the ones most likely to sink you.

The quick version

  • Managing up is consciously working with your boss to get the best result for you, them and the organisation, not flattery, but understanding their goals, pressures and working style so you can be genuinely useful.
  • Managing down is the familiar part: directing, developing and trusting the people who report to you, though even here the relationship is two-way, not one of command.
  • Managing across is getting things done through peers who don't report to you, using influence and exchange rather than authority you don't have.
  • The thread running through all three is mutual dependence: each relationship is a two-way deal, and the leaders who thrive treat all three directions as relationships to be managed, not taken for granted.

The idea in depth: every direction is mutual dependence

The phrase that unlocks this whole topic comes from John Gabarro and John Kotter's "Managing Your Boss" (Harvard Business Review, 1980, reissued as a Classic in 2005). Their argument was quietly radical for its time: the relationship with your boss is one of mutual dependence between two fallible human beings. You depend on your boss for direction, priorities, resources and a connection to the rest of the company; your boss depends on you for reliability, honesty and the cooperation that lets them do their own job. Treating that relationship as one-directional, the boss commands, you comply, wastes the half of it you actually control.

So the move is unglamorous and specific: invest time in understanding your boss as a person with their own goals, pressures, strengths, blind spots and preferred way of working, then adapt your own style to close the gap. Do they want detail or headlines? To read or to talk? To be surprised in the meeting or briefed before it? Gabarro and Kotter define managing up plainly as "the process of consciously working with your superior to obtain the best possible results for you, your boss, and the company." That word consciously is the whole point: most people manage their boss by accident, and badly.

The relationship with your boss is mutual dependence between two fallible human beings, and you control half of it.

The same logic runs downward, which is where managers most often forget it. You hold formal authority over your team, but authority is not the same as influence, and a team that complies is not a team that commits. The healthiest downward relationships are still exchanges, you provide clarity, air cover, growth and trust; they provide effort, candour and results. The shift here is from controlling to enabling: setting a clear direction and the boundaries of a decision, then handing real ownership across rather than hovering. (The mechanics of doing that well, how much to delegate, and to whom, are a topic in their own right; see role & expectation clarity.)

An honest limitation. "Manage your boss" is easy advice to give and miserable to follow when the boss is the problem. Gabarro and Kotter were describing an ordinary, well-meaning superior, not an abusive or chaotic one. The framework helps you take responsibility for the half of the relationship you control; it does not oblige you to absorb genuinely toxic behaviour, and it is not a substitute for HR, a transfer, or the door. Read it as "do your part well," not "it's always on you."

Managing across: influence without authority

The hardest direction is sideways. As organisations flatten and work crosses functions, more of what you need sits with peers who have their own bosses, their own priorities, and no reason to drop them for you. Herminia Ibarra, in Act Like a Leader, Think Like a Leader (Harvard Business Review Press, 2015), makes the case bluntly: lateral relationships with people outside your immediate area become more critical, not less, as you take on broader work, they are how you learn what's coming, sell your ideas, and compete for scarce resources.

The most usable model for the sideways problem is Allan Cohen and David Bradford's Influence Without Authority and its idea of currencies of exchange. Their premise: when you can't order someone, you trade. You work out what the other person values, a "currency", and offer something of value in return for the cooperation you need. Cohen and Bradford group the things people commonly value into a handful of currencies: inspiration-related (a sense the work matters), task-related (resources, help getting the job done), position-related (recognition, visibility, reputation), relationship-related (belonging, support) and personal-related (gratitude, autonomy, a sense of being valued).

flowchart TD
  A(["You need something
from a peer"]) --> B(["What does THEY value?
(their currency)"])
  B --> C(["Inspiration · Task
Position · Relationship · Personal"])
  C --> D(["Offer something of value
in return, now or later"])
  D --> E(["Cooperation earned,
relationship banked"])
  A -.->|"skip the exchange"| F(["Demand it →
resistance, no leverage"])
Cohen & Bradford's exchange model, influence sideways is a trade, not an order. Leaders Loop

So the move, the next time you hit a wall with a peer team, is to stop pushing your request and start asking a different question: what does this person need that I can provide? Maybe the analyst whose time you want is desperate for visibility with leadership, and you can give them a moment in front of your director. Maybe the engineering lead is buried, and the most influential thing you can do is take a meeting off their plate. Influence is built before you need it, in the currency the other person actually values, which is rarely the one you'd have guessed.

An honest limitation. The exchange model can curdle into transactional score-keeping if you treat every interaction as a trade. Cohen and Bradford are clear that the strongest currency is often a genuine, trusting relationship, and trust dies the moment people feel they're being worked. Use the currencies idea to widen your sense of what others value, not to turn every favour into an invoice.

flowchart LR
  U(["UP, your boss
direction, priorities,
resources, cover"]) --> Y(["YOU"])
  Y --> U
  Y --> D(["DOWN, your team
clarity, trust,
growth, air cover"])
  D --> Y
  A(["ACROSS, your peers
dependencies, budget,
information"]) --> Y
  Y --> A
Each arrow goes both ways: leadership in three directions, all of them mutual dependence. Leaders Loop

A worked example

Take a product manager, call her Dana, trying to ship a feature that needs design time she doesn't own, sign-off from a director who's hard to pin down, and data from an analytics team that reports elsewhere. (Illustrative scenario; not a real person or company.) Her instinct is to push: chase the director, escalate the design request, file a ticket with analytics and wait.

Played through the three directions, the better moves look different. Up: Dana knows her director reads nothing longer than a paragraph and hates being surprised in steering meetings, so she sends a three-line note the day before with the one decision she needs, framed against the goal he's accountable for. That's managing up: adapting to his style and his pressures rather than resenting them. Across: the design lead doesn't report to her and is swamped, so instead of escalating she asks what would make her request easy to say yes to, and offers, in exchange, to write the user-research brief the designer has been dreading. A task-currency trade. Down: rather than micromanaging her two engineers through the ambiguity, she gives them the goal and the constraints and lets them choose the approach, enabling, not directing.

Same feature, same lack of formal power over most of the people involved. The difference is that Dana stopped treating three relationships as obstacles and started treating them as relationships, each with its own currency, its own style, its own two-way deal.

Frequently asked questions

Isn't "managing up" just sucking up to the boss?

No, and the distinction matters. Flattery serves you; managing up serves the shared result. Gabarro and Kotter's framing is explicitly about getting the best outcome for you, your boss and the company by understanding what your boss is genuinely under pressure to deliver and helping them get there. If your "managing up" only makes the boss feel good and changes nothing about the work, it's flattery. If it makes their job and yours measurably easier, it's the real thing.

How is managing down different from just being the manager?

Being the manager gives you authority; managing down well means using less of it than you're entitled to. The trap is assuming formal power is enough, that people will commit because you can direct them. They'll comply, but commitment comes from clarity, trust and ownership, which you have to give rather than command. Managing down is the deliberate version of the role, where you treat the relationship as a two-way exchange rather than a chain of command.

Why is managing across so much harder than the other two?

Because you have no formal lever to pull. Up and down both run along reporting lines; across cuts against them. Your peers answer to someone else, optimise for their own goals, and owe you nothing by default. That's exactly why the exchange model matters, without authority, the only durable currency is value you offer in return, and trust you've banked before you needed it.

What if I genuinely don't have time to manage in three directions?

You don't have time not to. The hours lost to a blocked dependency, a surprised boss, or a team that's guessing at priorities dwarf the time it takes to send a one-paragraph pre-brief or ask a peer what they need. Managing in three directions isn't extra work bolted onto the job, done right, it's the thing that stops the job grinding to a halt.

Where do I start if I'm new to a role?

Start up and across, not down. Most new managers over-invest in their team in the first month and under-invest in the boss and the peers, then hit a wall when they need a budget or a dependency they never built a relationship for. In your first weeks, learn how your boss likes to work and what they're measured on, and meet the peer teams you'll depend on before you depend on them.

Related in the Toolkit

Managing in three directions starts with knowing who matters and how they sit on the power/interest map, and much of the up-and-across work is really the craft of securing buy-in from people you can't simply instruct.

Where to go next