High Output Management

The problem it starts from

In the early 1980s Andy Grove was Intel’s president, running a company that made some of the most demanding products on earth, and he noticed that the people he promoted into management were given almost no idea what the job was. Engineers knew what their output was: a working design, a yield figure, a shipped part. Managers spent their days in meetings and conversations and could not say what any of it produced. Many concluded, privately, that management was overhead and that their real work was whatever technical problem they still held on to.

Grove wrote High Output Management to give those people an answer. It was published in 1983, sold modestly, and became a cult book in Silicon Valley decades later, helped by founders who found that nothing else explained the job so plainly. A 2015 edition carries a foreword by Ben Horowitz, who has said the book shaped his own writing on management.

The practices

The book opens in a kitchen. Grove imagines a small operation that must serve a breakfast of a boiled egg, buttered toast and coffee, hot and at the same time, to a stream of customers. From that one image he draws most of the language of production: the limiting step that sets the pace for everything else (the egg, because it takes longest), the choice between building to order and keeping stock, where to inspect for defects, and why the cheapest place to catch a problem is at the lowest-value stage. Only then does he say that a manager’s work runs on the same principles, and the reader already has the vocabulary to follow.

Pair your indicators. Any single measure distorts behaviour, so Grove put each one beside a counter-measure: output with quality, inventory levels with the incidence of shortages. The pair keeps people from hitting a number by quietly damaging what the number was meant to protect. Long before “Goodhart’s law” became a common phrase in technology companies, this was a working defence against it.

Spend your time where leverage is highest. Grove’s central formula is that a manager’s output equals the output of the organisational units under their supervision or influence. It follows that the manager’s own activities should be judged by how much they move that larger output. Some activities have enormous leverage: a short conversation that unblocks a team, a decision made on time, a piece of information passed to the person who needed it. Others, such as doing a subordinate’s work for them, have very little, however satisfying they feel.

Hold one-on-ones, and let them belong to the subordinate. The regular meeting between manager and report is, in Grove’s design, the report’s meeting: they set the agenda, bring the problems and leave with something learned. How often it happens should depend on how experienced that person is at the particular task in hand.

Match your style to task-relevant maturity. This is the book’s most practical idea. A person can be highly mature at one task and a novice at another, so the right amount of supervision depends on the task, not the person’s seniority. Close structure suits someone new to the work; agreement on objectives and occasional monitoring suits someone who has done it many times. The common failures are to micromanage the expert and to abandon the newcomer, and both come from treating maturity as a trait of the person.

Make decisions in three stages. Free discussion, in which every view is aired; a clear decision, taken by whoever is responsible; and full support afterwards, including from those who disagreed. Grove was frank that peer groups find this hard, because colleagues of equal rank tend to wait for consensus that never arrives.

Set objectives, and say how you will know they were met. Intel’s management by objectives asked two things: where do I want to go, and how will I pace myself to see whether I am getting there. The second part, the key results, had to be concrete enough to check. John Doerr carried the practice from Intel to Google in 1999, and it spread from there as OKRs.

Train your people yourself. Grove argued that training is one of the highest-leverage things a manager can do and should not be left to a separate department, because only the manager knows what the team needs to learn.

The idea underneath

Beneath the practices is a refusal to treat management as a mystery. Grove brought the habits of a process engineer to a role that most writing of the time described in terms of charisma, character or inspiration. If output can be defined, it can be measured; if it can be measured, the process that produces it can be studied and improved. That is why the book moves so easily between factory floors and meeting rooms. Both are systems that turn inputs into outputs, with limiting steps, inventories and defects.

The second idea is humane in a way the engineering vocabulary can hide. Because a manager’s output is other people’s output, the manager’s real raw material is the motivation and ability of the people in the team. Grove treats performance reviews, training and one-on-ones as central work for that reason. The book’s sporting analogy makes the same point: a good coach does not play the game but is judged entirely by how the team plays.

Since it was published

The book’s ideas have outlived the industrial setting it describes. OKRs are now used far beyond technology, often in a form looser than Grove’s, with key results that cannot actually be checked. Remote and hybrid work have made his one-on-ones more important rather than less, since the hallway conversations that once carried information have thinned out. His insistence on paired indicators reads well alongside the idea page When a Measure Becomes a Target, which describes what happens when a single measure becomes the target.

A first step

Take your last full working week and list what you did. Beside each item, write which team’s output it changed and roughly how much. The exercise is uncomfortable, because many hours turn out to have touched nothing. Then pick one person on your team and one task they are doing now, and ask honestly how much experience they have with that specific task. Adjust how closely you follow it, in whichever direction the answer points.