Personalities
Sheryl Sandberg: What Leaders Can Learn From Scaling Operations And Influence
Growth creates a peculiar leadership temptation. Once an organisation has found momentum, it becomes easy to confuse the visible expansion with a reliable operating system.
Growth creates a peculiar leadership temptation. Once an organisation has found momentum, it becomes easy to confuse the visible expansion with a reliable operating system. More customers, more hiring, more markets and more attention can make a company look strong while its decisions, responsibilities and safeguards are becoming harder to see.
Sheryl Sandberg’s move to Facebook in 2008 provides a documented entry point into that tension. Facebook’s announcement said she would join as chief operating officer to help scale operations and expand globally, with responsibility across sales, marketing, business development, human resources, public policy, privacy and communications. It is a large mandate, not evidence that any individual is a universal leadership model. It does, however, make a useful question concrete: how should a leader organise influence when the work of scaling crosses functions and the consequences travel beyond the leadership team?
The answer is not “centralise everything” or “be more influential.” It is to make ownership, information and challenge legible enough that growth does not turn ordinary decisions into hidden risks.
Start with the decision in front of you
Take a familiar moment: a growing company launches in a new market. Product wants speed. Sales sees revenue. Legal sees a different regulatory environment. Support sees a predictable flood of confused customers. Public policy sees reputational exposure. None of these functions is necessarily obstructing the others. They are looking at different consequences of the same move.
The operational problem begins when no one is accountable for connecting those views before the launch. A leader may call that “empowerment,” but it is usually abandonment disguised as autonomy. Coordination is not a tax on growth when it makes unavoidable trade-offs visible early enough to act on them.
Meta’s 2008 announcement is useful historical context because it specifies a cross-functional remit rather than a vague aura of leadership. The announcement lists the functions and the stated purpose of scaling operations. The appropriate inference is not that copying a title or personality will scale another organisation. It is that scaling has to become somebody’s explicit work.
What the evidence can and cannot tell you
An employer announcement establishes what the company said Sandberg’s role would be. It does not establish that every subsequent decision was right, that a particular leadership philosophy caused an outcome, or that the organisation’s public record can be reduced to one executive. Large companies are systems of many people, incentives, technical choices, external conditions and accountability mechanisms.
That limitation is not an academic footnote. It protects teams from “leader biography as proof.” If a company grows, observers often retrospectively call every visible practice strategic. If it falters, they retrospectively call the same practice a flaw. A reader who wants a useful lesson needs a closer unit of analysis: a decision, a handoff, an information gap, an incentive or a check that could have changed the result.
Sandberg’s public work on workplace leadership is also not a universal answer to structural inequality. Lean In describes itself as an organisation offering programs on leadership development and confronting bias, while its current research describes barriers that are not solved by individual confidence alone. Its description is a reasonable source for what the organisation offers, not proof that one approach will fit every workplace. That distinction matters because influence is never only a matter of speaking up; organisations distribute opportunity, recognition and risk through systems.
The distinction that changes the next move
The central distinction is between having influence and building accountable influence.
Having influence means that people listen, resources move, and decisions follow. It can be useful and sometimes necessary. Building accountable influence means that the people affected by a decision can identify who owns it, what information was considered, where disagreement can be raised, and when the decision will be revisited.
The difference is clearest under pressure. A senior executive can clear a path by overriding a function. Sometimes the urgency justifies it. But if the override becomes the normal operating system, important knowledge stays silent or arrives too late. Teams learn that the real route to progress is access to the powerful person, not a process that makes the issue visible.
Accountable influence is slower only when a team mistakes hidden work for fast work. It uses a small number of repeatable commitments: a named decision owner, explicit input owners, a record of assumptions, a date for reassessment and a route for escalating material risk. These are not bureaucratic decorations. They are how an organisation remains capable of learning when the person with the loudest voice is absent.
A practical way to test your interpretation
Choose one decision that crosses at least three functions: a launch, a pricing change, a major vendor choice, a hiring plan or a change to customer terms. Create a one-page operating view.
Start with the outcome and deadline. Name the one person who will decide, not merely convene. Name the functions whose information could materially change the decision and ask each for the specific risk, dependency or opportunity they see. Write assumptions in plain language. Identify which one would be costly if false.
Then ask two questions that often reveal whether influence is accountable. First: if the decision goes wrong, who will carry the consequence and have they been heard? Second: what evidence, by what date, would cause the decision owner to pause, change course or escalate?
The goal is not consensus. Consensus can hide responsibility just as effectively as executive fiat. The goal is a decision that is intelligible: people know the trade-off, the owner and the recovery route. In a fast-moving context, that clarity can reduce rework because later objections are not surprises.
For high-impact decisions—those affecting safety, privacy, employment, finance, legal obligations or vulnerable users—raise the evidence standard. A tidy operating note is not specialist advice. Bring in qualified expertise and ensure the decision respects the applicable rules. Scaling operations does not make these boundaries less important; it increases the number of people who can be harmed when they are ignored.
Make the operating view usable after the meeting as well. A decision owner should be able to say which parts are settled, which are conditional and which function has authority to stop work if a named threshold is crossed. “Consult legal” or “check with support” is not an owner or a threshold. A more useful statement is: “Privacy will review the proposed data flow before implementation; no customer data will be transferred until that review is complete.” The exact safeguards will vary by domain, but plain language prevents a dependency from being mistaken for a completed check.
It is also worth distinguishing speed from compression. A fast decision gives the necessary people a bounded window to surface material information, chooses with what is known, and makes the review date real. A compressed decision eliminates the window and describes the later surprise as an execution problem. If the deadline cannot move, change the scope, sequence the rollout, or make the uncertainty explicit. Those choices may still be difficult, but they keep a commercial target from silently becoming everyone else’s unchosen risk.
Finally, test whether the process works for someone without personal access to the decision maker. Ask a person close to implementation where they would raise a material concern, what response they would expect, and whether they could tell the difference between a request for input and a final instruction. If the answer relies on knowing whom to message privately, the operating system is fragile. The corrective may be as small as a named forum, a shared record and a response expectation; it does not require turning every ordinary decision into a committee.
Common mistakes that create false certainty
The first mistake is confusing a detailed dashboard with shared understanding. Metrics can show movement without revealing who owns a bottleneck or whose work has been shifted elsewhere. Pair numbers with a clear account of the decision they inform.
The second is over-centralisation. A leader who approves every exception may feel informed while creating a queue and teaching people not to exercise judgment. The repair is not total decentralisation. It is to define which decisions can be made close to the work, which require cross-functional review, and which need executive accountability.
The third is treating dissent as a communication failure. A legal, safety or frontline objection may be inconvenient; it may also be the information that keeps a growth plan from becoming an avoidable harm. Ask the objector to state the condition under which they could support the decision. Ask the sponsor to state the evidence that would make them reconsider. This moves the conversation from status to testable claims.
The fourth is celebrating personal influence while ignoring unequal access to it. Who gets invited into a decision, whose evidence is treated as strategic, and who can afford to disagree are organisational design questions. Individual courage can matter, but it cannot substitute for fairer systems of sponsorship, promotion, feedback and protection from retaliation.
The daily version of operational scaling is less dramatic than a reorganisation chart. It is the recurring meeting where the same decision returns because no one knows what was decided. It is the customer-facing team that learns of a change from customers. It is the risk team asked for approval after the commercial commitment has been announced. These are signs that information is moving through status rather than through a dependable operating path.
Start by mapping one handoff, not the whole company. Choose a request that regularly crosses teams: a new feature, a customer exception, an incident response or a vendor approval. Follow it from request to decision to execution to feedback. Where does it wait? Who is required to interpret information that they do not receive? Which responsibility is shared so broadly that no person feels able to close it? The answer may be a role gap, a conflicting incentive, an unclear threshold or a missing forum. It is rarely solved by asking everyone to “communicate more.”
Influence is especially fragile when a team has mixed incentives. Sales may have a legitimate reason to seek speed; security may have a legitimate reason to require review. Calling one side commercial and the other obstructive produces an argument about identity. A better operating question is: what decision would let us know whether the revenue opportunity is real while containing the harm if the security concern is also real? A phased rollout, a constrained pilot, a contract condition or an independent review may be available. The right choice depends on the domain, but the shift from status to trade-off is transferable.
This framing also clarifies the difference between consultation and delegation. Consultation means a decision owner seeks relevant knowledge while retaining responsibility. Delegation means another person has defined authority to decide within boundaries. Teams get into trouble when they call consultation delegation: a junior lead is apparently empowered, but a senior executive repeatedly reverses the choice. They also get into trouble when consultation has no endpoint and nobody will decide. Name the mode before the work starts.
Records can support this without becoming performative. A short decision record should not turn every choice into a report. Reserve it for decisions with cross-functional impact, material cost or a meaningful chance of reversal. Use plain language. What are we choosing? Why now? What did we consider? What would change the course? Who needs to know? A record that cannot be read by the people implementing the work is not operational clarity; it is archival comfort.
Leadership influence becomes more trustworthy when it includes a recovery route. A plan that assumes success has no place to put bad news. Before launch, agree on the signal that would require a pause, the person authorised to trigger it and the way customers or colleagues will be supported if it happens. This is not pessimism. It is respect for the fact that scale amplifies both useful systems and unattended errors.
The public record around any well-known company is necessarily incomplete from an outsider’s view, and it can be tempting to turn it into a moral fable. Resist that. The value for a reader is not an adjudication of an individual’s whole career. It is a clearer lens for their own operating environment: power should make responsibility easier to locate, and growth should make important information easier—not harder—to surface.
Choose the smallest responsible next step
Choose one upcoming cross-functional decision. Before the next meeting, write the owner, the affected functions, the costly assumption, the escalation route and the date at which the decision will be reviewed. Send it to the people who have to live with the result and ask what critical information is absent.
When the responses arrive, do not turn the exercise into a search for agreement. Look for the information that changes the trade-off. Perhaps the commercial deadline is fixed but the proposed scope is not. Perhaps the privacy risk can be contained with a different data flow. Perhaps a frontline team has already seen the failure mode in another form. These details make an operating decision more intelligent because they turn “alignment” into a practical description of what must be held together.
After the decision, close the loop. Tell contributors what was chosen, what was not chosen and why. If the outcome differs from the expectation, return to the assumptions rather than immediately searching for someone to blame. The habit is simple, but it changes the social meaning of escalation: people can raise a risk because it may improve the work, not because they must win an argument.
There is a useful boundary here. A process cannot make every organisation fair, safe or wise, and it cannot absolve senior leaders of responsibility for the choices they authorise. It can make responsibility harder to evade. For readers leading a small team, that may mean one shared decision note and a regular review. For a larger organisation, it may mean clearer mandates, specialist oversight and protected routes for difficult information. The scale differs; the principle remains: influence earns trust when it exposes rather than conceals its consequences.
The same principle applies to communication after a decision. Announcing a result is not enough when implementation requires people to alter priorities, scripts, systems or customer expectations. Explain the operational consequence in the language of each group. What should change tomorrow? What remains unchanged? Who can answer a question when the written decision does not fit the case? A useful leader does not create dependency by being the only interpreter. They create enough shared context that sound judgment can travel.
It is also worth checking for silent concentration of risk. A decision can appear well owned because a senior person has signed it, while the frontline team absorbs the errors, rework and customer frustration without authority to prevent them. Ask those closest to the work what happens when the plan meets an exception. If the answer is “we improvise and hope it is not noticed,” the organisation has found a scaling problem before it becomes a public one.
This makes the leadership lesson practical without pretending it is effortless. Clear ownership requires time. Listening to inconvenient information requires humility. Revisiting a decision can threaten status. But the alternative is often costlier: a culture in which people learn to route around responsibility until a consequence forces attention. Scale needs influence; responsible scale needs influence that remains open to correction.
Start with one decision rather than a new management philosophy. If people can see the owner, the trade-off and the recovery route, they can contribute with more confidence. That is how operational influence becomes a shared capability rather than a personal performance.
It is a capability teams can practice, inspect and improve together.
That small practice will not turn a company into a global organisation, and it should not pretend to. It can make one decision more accountable. That is a more grounded lesson from a documented scaling role than trying to imitate a public figure: make influence visible enough that it can be challenged, improved and responsibly used.
