पाठशाला Pathshala · संचालन Sanchālan, Operations · Lesson 24 · Scale

The COO decision: do you need one and what should they own

There is no single COO job. There are at least three, and the failures come from hiring for one and describing another. Decide which you need, then write down what they own.

Pathshala, The Founder Library · 11 October 2026 · 7 min read

Two pilots seen from behind at the controls of an airliner during take-off. — Photograph: K · Pexels

Somewhere past fifty people most founders hear the same advice, from an investor, a peer or their own exhausted reflection: you need a COO. Sometimes it is right. More often it names a solution before anyone has named the problem, and a senior hire arrives into a role that was never defined and leaves a year later with nobody quite sure what went wrong.

This lesson sets out what the COO role actually is, the three different jobs that share the title, the symptoms that point to each, a decision tree to walk, the written charter that makes the role work, and how to hire and review it. It is written for founders of companies between fifty and two hundred people.

There is no single COO job

The most useful thing ever written on the role is Nate Bennett and Stephen Miles’s Second in Command, in the Harvard Business Review of May 2006. Their starting point is in the title: of all the senior roles it is the one least understood, and they identify seven distinct reasons companies create it: to implement the CEO’s strategy, to lead a specific initiative such as a turnaround, to mentor a young or inexperienced CEO, to complement the CEO’s strengths or offset their weaknesses, to act as a partner to the CEO, to test a possible successor, and to keep a valuable executive who might otherwise leave for a rival.

Seven reasons means seven jobs. A COO hired to offset a founder’s weakness in operations and a COO hired as a successor-in-waiting need different people, different authority and different measures. Most failures come from hiring for one reason and describing the role as another, or from never saying which reason it is.

Three roles that get called COO

In a company of fifty to two hundred people, the useful distinction is between three roles. The operating COO owns a group of functions outright, typically operations, customer support, finance and sometimes people, and the heads of those functions report to them. The founder keeps product, the largest customers, fundraising and the board. This is the complement Bennett and Miles describe, and it is the role most founders mean when they say COO.

The chief of staff owns no function and has nobody reporting to them. They extend the founder’s reach: running the operating cadence, preparing the leadership meeting, chasing decisions to completion, owning cross-team projects that have no home, and telling the founder what the team will not. The role is cheaper, faster to hire and easier to undo. It is also the right answer more often than founders expect, because the problem is frequently not that the founder has too much to decide but that too much of what was decided never happens.

The head of a function with a grander title. A company whose supply chain, delivery network or support operation is failing needs someone who has run that operation at the next scale. Calling that person COO adds nothing except a promise of authority they will not have, and a reason for the other heads of function to resent them.

The symptoms, read honestly

Each role answers a different symptom. Decisions queue at the founder because ten or twelve people report to them and every cross-functional question needs them in the room: that is a span problem, and an operating COO who takes four or five of those reports solves it. Things are decided and then nothing happens, follow-ups drop, the weekly review produces actions that nobody chases: that is a reach problem, and a chief of staff solves it. One part of the business is broken and everything else is fine: that is a function problem, and a head of function solves it. An investor says hire a COO: that is not a symptom, and the right response is to ask which of the first three they are seeing.

There is a fourth symptom that no hire solves. If the founder is in every meeting because the company has no operating rhythm, no weekly metrics review, no monthly close, no written decisions, then a senior operator walks into chaos and will be blamed for it. Fix the [operating cadence](/library/founders-operating-cadence) first. It takes a quarter, costs nothing, and makes any later hire more likely to work.

Writing down what they own

Whatever the tree returns, write a charter before the search starts. One page, four headings. Functions: which teams report to the hire, by name. Decisions: the decisions they make alone, the ones they make with the founder, and the ones that stay with the founder, with an example of each; the founder must be able to read the first list and accept being overruled in it. Metrics: three to five numbers they answer for, taken from the [weekly page](/library/weekly-metrics-review-one-page-one-hour), such as gross margin, delivery time, support resolution time or the close date. What stays with the CEO: usually product, fundraising, the board, the largest customers and the final say on senior hires.

Aviation headsets hanging in an empty aircraft cockpit. — The second seat is defined before anyone sits in it. A COO charter does the same job on one page. Photograph: Luis Quintero · Pexels

The decisions list is the part founders find hardest and the part that decides whether the role works. A COO who owns support but whose every policy change is reviewed by the founder owns nothing. Amazon’s distinction between reversible and irreversible decisions, set out in Jeff Bezos’s 2016 letter to shareholders, helps here: most decisions are two-way doors and can be made by the person closest to them with a light process. Put the two-way doors in the COO’s list. Keep the one-way doors, the ones that cannot easily be undone, in the shared list.

The charter matters even more when the COO is a co-founder. Many founding teams give one founder the title at incorporation, when the company has five people and nothing to operate, and the title then grows by accretion into whatever that founder happens to be doing. At fifty people the CEO and the co-founder COO should sit down with a blank page and write the charter as if hiring from outside: which functions, which decisions, which numbers. If the result is a role the co-founder does not want or is not suited to, it is far better to find that out in a written conversation between partners than in a slow year of overlapping authority that the whole leadership team can see.

A COO is not a person. It is a written division of the company between two people, and it works only if both have signed it.

Hiring for it

Ben Horowitz’s advice on hiring an executive for a job you have never done applies directly. Know what you want, and the best way to know is to act in the role for a few weeks. Write down the strengths you need and the weaknesses you will tolerate. Build interview questions that test for those strengths. Run the reference checks yourself. And make the final call alone, because consensus on executives drifts towards the candidate with the fewest weaknesses rather than the one with the strengths the job needs.

Be careful with candidates from large companies. Horowitz’s companion essay names the two mismatches. The rhythm: a large-company executive’s day is full of incoming requests, while in a startup nothing happens unless the executive starts it. And the skills: running an existing machine rewards prioritisation and process improvement, while building one demands hiring from nothing and creating processes that do not yet exist. Ask every candidate what they will do in their first thirty days. The answer tells you which kind of operator they are.

In India the search has one more test worth adding. A COO who will own operations in a company selling to Indian SMEs or running a field network must have run that kind of operation here, with its working-capital cycles, its compliance calendar and its distances. A strong operator from a global software company may not have, and the charter should say which experience is required.

The ninety-day review

On the day they join: give them the charter, agree three objectives for the first ninety days with a measure each, and announce to the company which teams now report to them and which decisions are theirs. Weekly for the first quarter: a one-hour one-on-one with the founder, with the decisions list on the table, adding to it as trust builds. At thirty days: ask both of you whether the hire is up to speed; Horowitz’s view is that a new executive who is not should go, and a founder who waits six months to act on that loses the quarter and the team’s confidence. At ninety days: review the three objectives, rewrite the charter with what was learned, and confirm the decisions list in writing. Once a year: review the charter as a pair, because as the company grows the division between founder and COO will need to move.

The tree and the thresholds are a starting point for a conversation between co-founders and the board, not a verdict. The sources below are short and worth reading in full before the search begins.