This is a user-friendly guide to the art and science of management from Andrew S. Grove, the president of America's leading manufacturer of computer chips. Combining conceptual elegance with a practical understanding of the real-life scenarios that managers encounter every day, High Output Management is one of those rare books that have the power to revolutionize the way we work.

Any errors, omissions, or misrepresentations are mine.


Rules of the modern business environment: (1) everything happens faster; (2) anything that can be done will be done, if not by you, then by someone else. Try to anticipate the unexpected, and when it happens, double your efforts to make order from the disorder. Let chaos reign, then rein in chaos.

The output of a manager is the output of the teams under his/her supervision or influence. High managerial productivity involves choosing to perform tasks that possess high managerial leverage. Try to elicit peak performance from all individuals on your team, and plan the way a fire department does -- you cannot anticipate the next fire, but you can create an energetic and efficient team that is capable of responding to any event, anticipated or not.

Continually dedicate yourself to retaining your individual competitive advantage. Are you adding more value? Are you plugged in to what's happening around you? Are you trying new ideas, new techniques, and new technologies?


There are three fundamental types of production operations: process manufacturing (material changes), assembly (components put together to constitute a new entity), test (examination of its characteristics). Design your production flow by starting with the limiting (longest, most difficult, most sensitive, most expensive) step, then working backwards.

To run your operation well, you need a set of good indicators, or measurements. Choose indicators that are a physical, countable thing, and that measure output as well as activity. Focus each indicator on a specific operational goal. Pair indicators so both effect and counter-effect are measured (as indicators will naturally direct your attention and effort in one direction) — e.g. quantity vs. quality, or in software, completion date vs. capability. Leading indicators (measuring linearity or trends in measurements) can show what the future might look like.

All production flows increase value of a material as it moves through the process. Try to detect and fix any problem at the lowest-value stage possible. You can lower the cost of QA by performing monitoring instead of gated inspections, and by using variable inspections where test frequency is proportional to problem rate.

Slack should be designed into the system to mitigate unpredictability in forecasts. Slack in the form of inventory should be at the lowest-value stage to maximize production flexibility.

Productivity is the output divided by the labor required to generate the output. It can be increased by performing the work activities at a higher rate, or by increasing the leverage of the activities.

Managerial Leverage

A manager's output is not her individual work, but instead the output of her organization plus that of the neighboring organizations under her influence. She should shift energy and attention to activities what will most increase organizational output, e.g. moving to the point where her leverage is greatest.

Most useful information in information gathering comes from quick, casual verbal exchanges. Reports and plans have value, but often in the discipline involved in the preparation itself instead of the final output. Managers are also sources of information. A manager conveys his knowledge, objectives, priorities, and preferences to his own org and groups he influences, creating a shared corporate culture. Culture is indispensable as it creates deterministic behavior — team members will behave in a consistent fashion under similar conditions.

There are two types of decisions: the forward-looking sort where resources (manpower, money, capital) are allocated, and the decisions responding to a developing problem or a crisis. Your skill as a decision maker depends on how well you comprehend the facts and issues facing your business, which is why information gathering is so important.

Often, you do things designed to influence events slightly — nudging individuals or teams in a direction you would like. This is immensely important as it occurs far more often than making unambiguous decisions.

Nothing leads as well as example. Values and behavioral norms are transmitted most easily by doing and doing visibly — you must be a role model for people in your organization.

The most important resource you allocate from one day to next is your time. How you handle your time is the single most important aspect of being a role model and a leader. You should focus on increasing the leverage of your activities. High leverage can be achieved in one of three ways:

  1. When many people are affected by one manager — this type of leverage often depends on when it is performed, so keep timeliness in mind.
  2. When a person's activity or behavior over a long period of time is affected by a manager's brief, well-focused set of words or actions. This can be positive (e.g. performance reviews) or negative (being a depressed, waffling, or meddling manager). The lack of a decision is the same as a negative decision.
  3. When a large group's work is affected by an individual supplying a unique, key piece of knowledge or information.

If you have a choice, you should delegate activities you know best — monitoring the delegated task is the only practical way to ensure a result, and it is easier to monitor something with which you are familiar. Be sure to monitor at the lowest-added-value stage of the process: review rough drafts before they have spent time polishing them. Use a variable frequency in checking your subordinates' work — modify frequency based on their experience and prior performance with the task. You can delegate certain types of decisions too — just be sure to monitor your subordinates' decision-making process.

The other way to increase managerial output is by increasing the rate of performing work. This can be done by:

  1. Finding the limiting step, then scheduling other work around it on your calendar.
  2. Batching similar tasks.
  3. Forecasting the things you can to minimize fragmentation.
  4. Saying no at the outset (the lowest-value stage) to work beyond your capacity to handle.
  5. Allowing slack in your scheduling.
  6. Carrying an inventory of projects you need to do but don't need to finish right away — this will ensure you do not spend free time meddling in your subordinates' work.
  7. Trying to standardize what you do, but continuing to think critically about what you do and the approaches you use.

To maximize leverage without the risk of meddling or getting bogged down, you should have 6-8 subordinates or their equivalent.

Strive toward regularity — try to smooth out your workload as much as possible. To reduce uncontrolled interruptions, you can prepare standard responses for the most frequent ones, then applying batching (in the form of regular staff and one-on-one meetings) to move them into standard times.


A meeting is the medium through which managerial work is performed. Just make sure you use the time spent in them as efficiently as possible.

In process-oriented meetings, information is exchanged. These often take place on a regularly scheduled basis.

One-on-ones are for mutual teaching and exchange of information. The supervisor talks about specific problems and situations to teach the subordinate his skills and know-how; the subordinate provides the supervisor with detailed information about what he is doing and what he is concerned about. These should last an hour at a minimum, and the frequency should be higher for subordinates with less experience, or when things change faster in his job area. Taking notes is valuable for both the supervisor and the subordinate.

The agenda and tone should be set by the subordinate, but should generally focus on indicators that signal trouble, anything important since the last meeting, any problems or potential problems in general (especially those that preoccupy or nag the subordinate), and any heart-to-heart issues. The supervisor should facilitate the subordinate's expression of what is going on — often, this can be done by getting the subordinate to say all he wants about a subject, then asking one more question.

Staff meetings are comprised of a supervisor and all of her subordinates. They are an opportunity for interaction between peers and create learning opportunities for the supervisor in any exchange/confrontation that develops. The time should be spent discussing anything that affects more than two of the people present — if it degenerates into a conversation between two people about a problem affecting only them, they should be suggested to continue their conversation later. The agenda should be fairly structured, but have an "open session" for staff to bring up anything they want. The supervisor should be active mainly to keep the discussion on track, as a leader, observer, expediter, questioner, and decision-maker.

Operation reviews are used as a medium of interaction between people who don't otherwise have much opportunity to deal with one another; they keep teaching and learning going on between employees several organization levels apart. The agenda should include formal presentations, and it should include a senior supervisor (to ask questions and act as a role model), an organizing manager who is in charge of time-keeping and reviewing/helping with the presentations, the presenters themselves, and an audience (for immediate feedback).

Mission-oriented meetings frequently produce a decision, and due to their nature often can't be scheduled. The chairman of the meeting (usually who calls it) should have a clear understanding of the meeting's objective, get commitments for attendance, maintain discipline, and finally send minutes about the discussion, the decisions made, and the actions to be taken. If process-oriented meetings are done correctly, these should be rare (never more than 25% of anyone's time).


In technology companies, new hires hold power because of their more recent knowledge, and veteran managers have power due to their position. Middle managers are the link between the two.

Steps to decision making:

  1. Free discussion — everyone voices their opinions as equals; all points of view and all aspects of an issue are openly welcomed and debated.
  2. A clear decision must be made — the greater the disagreement about the issue, the more important this is.
  3. Full support (commitment to backing) must be obtained from everyone involved.

Decisions should be made at the lowest competent level, where there is a balance between technical knowledge and experience in ups and downs in applying that knowledge.

The peer-group syndrome can occur if there is no clear leader — people will talk in circles and it will go nowhere. If no supervisor can be present, the person who has the most at stake should take charge. Don't let people be paralyzed by the fear of sounding dumb or the fear of being overruled. When all points of view, facts, opinions, and judgments have been aired, the leader may have to make a decision (but in this case it will be legitimate and without prejudice).

To ensure that the output (the decision) is achieved, the manager should be clear about:

  1. What decision needs to be made?
  2. When does it have to be made?
  3. Who will decide? (If there is more than one interest group, ensure they have roughly equal representation.)
  4. Who will need to be consulted prior to making the decision?
  5. Who will ratify or veto the decision?
  6. Who will need to be informed of the decision?

If the final word has to be dramatically different from the expectations of people who participated in the decision-making process, make sure you help everyone accept and learn to live with it.


Your general planning process:

  1. Establish projected need or demand — what will the environment, your business, your organization demand from you, in the near future and a year from now?
  2. Establish your present status — what are you producing now, and when your projects in the pipeline are completed? Where will your business be if you do nothing differently?
  3. Compare steps 1 and 2 — what more (or less) do you need to do to produce what your environment will demand? What can you do? When do you need to do it?

The abstract summary of your actions is your strategy; what you'll do to implement the strategy is your tactics. Frequently, a strategy at one managerial level is the tactical concern of the next higher level.

Planning should produce tasks that need to be performed now in order to affect future events — today's gap shouldn't exist in the first place, and is a failure of planning sometime in the past. The output of the planning process is therefore the decisions made and the actions taken as a result of the process. Be sure to give yourself enough time between planning to judge the impact of the decisions you made.

Remember saying "yes" to something is saying "no" to something else — have the guts, honesty, and discipline to drop projects as well as initiate them.

Use management by objectives to ensure pace after planning: create objectives by asking where you want to go; then enumerate your key results to see if you are getting there. Choose few crucial objectives to ensure focus, and keep your key results specific so it is clear whether they are completed or not. Note a manager's objectives will be tied to his supervisor's — if the manager meets his objectives, his supervisor will meet his.

Hybrid Organizations

Management is not just a team game, but a game for a team of teams, where individual teams exist in mutually supportive relationships with each other.

There are two extremes: a mission-oriented, decentralized organization where each business unit (e.g. product area) pursues what it does with little tie-in to other units; or a functional, completely centralized organization where each functional unit (e.g. sales, manufacturing) is responsible for its specific role everywhere. The former enables speedy response to changes in specific business or product areas, but the latter enables taking advantage of economies of scale and leveraging expertise in each operational area on a greater scale.

Good management is a reconciliation of centralization and decentralization — a balancing act to get the best combination of responsiveness and leverage. To make hybrid organizations work, the resources of the functional unites must be allocated and delivered to meet the needs of the mission-oriented units.

This can be done through dual reporting, where a manager in a business unit reports to someone in both the mission-oriented organization (e.g. his division general manager) and functional organization (a peer group of managers in the same role, or a global director of that role). The resulting two- or multi-plane organization structures allow for maximum flexibility for the organization and maximum leverage for each individual.

Modes of Control

Free-market forces: goods and services are exchanged between two entities, each seeking only to enrich itself. Value is the only thing that matters, and no one needs to oversee the transaction. This doesn't work when the value of something is not easily defined, e.g. in a business when a group of people is needed to accomplish a task.

Contractual obligations: one party has generalized authority (monitoring, evaluation, and correction) over another one by a set of rules or a contract. This doesn't work when the environment changes more rapidly than rules can be changed, or when circumstances are so complicated a contract that covers all possibilities will be prohibitively complicated.

Cultural values: the interest of the larger group to which an individual belongs takes precedence over the interest of the individual himself. Everyone shares a common set of values, a common set of objectives, and a common set of methods. The role of management is to articulate the values, but more importantly lead by example.

The most appropriate mode of control depends on the nature of a person's motivation, and the nature of the environment (in terms of complexity, uncertainty, and ambiguity, or CUA) in which he works. The market mode works best with high self-interest and low CUA environments; the contractual mode with high group interest and low CUA environments; the cultural values mode with high group interest and high CUA environments. High self-interest and complex environments can only produce chaos.

New employees have high self-interest, so should be given structured jobs with low CUA; later, when he gains shared experience, he will be ready to tackle higher CUA tasks (a role which tends to pay more).

All three methods of control exist at different circumstances at work.


When a person is not doing a job, he is either not capable or not motivated. The single most important task of a manager is to elicit peak performance from her subordinates, and she does this through training and motivation respectively.

Maslow's theory of motivation: needs cause people to have drives which result in motivation. A need satisfied stops being a need and therefore a source of motivation, and so to maintain a high degree of motivation, you must keep some needs unsatisfied at all times.

Maslow's hierarchy of needs:

  1. Physiological needs: things money can buy, like food, clothing, and other necessities. Fear is tied to possible deprivation of these necessities.
  2. Security/safety needs: desire to protect oneself from slipping back into a state of being deprived of basic necessities.
  3. Social/affiliation needs: desire to belong to groups whose members possess something in common with themselves.
  4. Esteem/recognition needs: desire to keep up with or emulate someone.
  5. Self-actualization needs: the need to achieve one's utter personal best in a chosen field of endeavor.

When physiological, security/safety, social, or recognition needs are met, the motivation is extinguished — but self-actualization continues to motivate people to ever higher levels of performance. Once someone's source of motivation is self-actualization, his drive to perform has no limit.

Two inner forces can drive a person to use all of his capabilities: competence-driven people want job or task mastery, achievement-driven people are moved by an abstract need to achieve in all they do. Management needs to create an environment that fosters these forces, helping everyone strive for a level of achievement beyond his immediate grasp while valuing and emphasizing output.

Once in self-actualization mode, a person needs measures to gauge progress and achievement. The most important type of measure is performance feedback, such as performance indicators and milestones in a MBO system that are linked to performance of the individual, or a supervisor performance review. Note that fear is not a good motivator in self-actualization mode — you cannot stay in that mode if you are always worried about failure.

The role of manager is that of coach: he takes no personal credit for the success of his team (so his players trust him); he is tough on his team, being critical to get the best performance his team members can provide; and generally has played the game well in the past so understands it well.

A given managerial approach is not equally effective under all conditions — it depends on the task-relevant maturity (TRM) of both the manager and her subordinates. When TRM is low, the most effective approach is one with very precise and detailed instructions. As TRM grows, the style should shift to one more given to communication, emotional support, and encouragement; the manager should pay more attention to the individual rather than the task at hand. At high TRM, the manager's involvement should be kept at a minimum, and primarily consist of making sure the objectives toward which the subordinate is working are mutually agreed upon. Of course, regardless of TRM, the manager should always monitor a subordinate's work closely enough to avoid surprises.

The responsibility for teaching the subordinate must be assumed by the supervisor. Values should be imparted early on, so later the subordinate can be trusted to make decisions in the future the way the supervisor would. Once operational values are learned and TRM is high enough, the supervisor can delegate tasks to the subordinate, increasing his managerial leverage.

Managers tend to see themselves as delegating more than they really are — watch out for situations where a manager throws out suggestions to a subordinate who receives them as marching orders.

Performance Appraisal

Performance reviews are the single most important form of task-relevant feedback supervisors can provide. Their fundamental purpose is to improve the subordinate's performance, and are dedicated to finding the skill level of the subordinate (to determine which skills are missing and to remedy that lack) and to intensity the subordinate's motivation to create a higher performance curve for the same skill level.

Assessing performance

A supervisor should clarify in his own mind in advance what it is he expects from a subordinate, then attempt to judge whether he performed to expectations. You can characterize performance by external measures (the output; e.g. completing designs, meeting sales quotas) or by internal measures (what is being done to create output in the current period, and to set the stage for output in future periods).

Be aware of the tradeoff between long-term-oriented and short-term-oriented performance — an engineer on a project can be working to complete the project on schedule, but also working on a design method that will make it easier for others to design similar products in the future. Think about how much the future-oriented activity will pay back over time, and how much that is worth today.

Also be aware of the time offset of the subordinate's activity and the output that results from that activity — one's output during the review period may have all, something, or nothing to do with one's activities in the same period.

When reviewing a manager, you should judge both his performance and the performance of the group under his supervision. Is he adding value to his group — is he hiring new people; training the people he has; doing anything to improve team output in the future? The performance rating of a manager should not be higher than that of his organization.

Promoting someone is effectively creating a role model for others in the organization — when we promote our best, we are communicating that performance is what counts.

Delivering the assessment

Three L's: level, listen, leave yourself out. Be totally frank, make sure that you are truly heard and understood by watching the subordinate's response, and understand that the review is completely for and about your subordinate.

Your subordinate has only a finite capacity to deal with facts, so target only a few key areas that will improve his performance the most. This involves preparation — write down everything in your head on paper, then look for relationships between the items listed to create messages for the subordinate. Prioritize these messages and only deliver the most important ones.

If you are dealing with a performance problem, the subordinate may progress through stages of conflict resolution: (1) ignoring it, (2) denying it, (3) blaming others, (4) assuming responsibility, and (5) finding a solution. (4) is the biggest step; you should be okay with any outcome that includes a commitment to action, even if you don't necessarily agree it is the optimal solution.

No matter how stellar a person's performance level is, there is always room for improvement — use the 20/20 hindsight provided by the review to show anyone, even an ace, how she might have done better.

Asking the subordinate to evaluate the supervisor's performance can be a good idea, but only if it is clear the assessment has only advisory status — do not pretend the supervisor and subordinate are equals.

Other Managerial Tasks

Interviewing a potential employee

The purpose of the interview is to select a good performer, educate him as to who you and the company are, determine if a mutual match exists, and to sell him on the job.

The applicant should do 80% of the talking during the interview, and what he talks about should be your primary concern. Steer the discussion toward subjects familiar to both you and the candidate, focusing on the applicant himself, what he has done and why, what he would have done differently if he had to do it over, etc.

You are trying to judge the candidate's potential contribution using information in four categories: (1) technical knowledge specific to the role, (2) performance in an earlier job using his skills and technical knowledge, (3) reasons for discrepancies between what his capabilities and his performance, and (4) his set of operational values.

Best questions to ask:

When a subordinate decides to quit

When an employee who is dedicated and loyal feels his work is not appreciated, his decision to leave reflects on you — you have not done your job, and have failed as his manager.

When this happens, drop what you are doing, sit the employee down and ask him why he is quitting. Keep asking questions so the real issues will come out. Don't argue, lecture, or panic — convey by what you do he is important to you, and try to find out what is really troubling him.

Vigorously pursue every avenue available to you to keep him with your firm, even if means transferring him to another department. You owe it to your employer to save an employee for the company.

Compensation as task-relevant feedback

If the absolute amount of a raise in salary is important, the person is likely motivated by physiological or safety/security needs. If the relative amount of a raise (what he got compared to others) is important, the person is likely to be motivated by self-actualization — money here is a measure, not a necessity.

Be very sensitive toward the various money needs of our subordinates and show empathy toward them. But use money as a way to deliver task-relevant feedback — base a portion of compensation on an employee's performance (as a bonus); this percentage should rise with total compensation.

Compensation can be based on experience (increases with time with company), merit (performance only), or a compromise between the two. Merit-based compensation is the only way to encourage performance, but you must be honest and understand that if someone is first, someone else must be last.

An achiever will alternative between "meets requirements" and "exceeds requirements" throughout his career (being promoted upon reaching the latter), until he settles at a "meets requirements" level.


A manager is responsible for her organization's output, and is therefore responsible for training her employees. Training is one of the highest-leverage activities a manager can perform — even a 1% improvement in your subordinate's performance will result in the company gaining the equivalent of hundreds of hours of work.

Two types of training tasks: (1) teaching new members of the organization the skills needed to perform their jobs, and (2) teaching new ideas, principles, or skills to the present members of the organization.

Implementing training:

  1. Make a list of things your subordinates should be trained in. Ask them what they feel they need.
  2. Take an inventory of the teachers and instructions available to help deliver training on these items. Assign priorities to these items.
  3. Start unambitiously — develop a short course on the most urgent subject. Set a schedule for your course, with deadlines, and commit yourself to it. Develop the second lecture after you have given the first, so you have a better feel for it.
  4. After the course, ask for anonymous critiques from employees in your class.