Skip to content
Question Vault?
Free to readNo accountNo email wallNo invented statisticsNo ads on medical, legal or end-of-life pagesCopy or print any set and take it with you
03 · Professional & Academic

Good Questions to Ask a COO

Twenty questions for someone who has to form a judgment about an operation they do not run: an investor or board member before a meeting, a consultant scoping the work, an acquirer, or a senior candidate deciding whether the job is doable. They are deliberately specific, because a COO can answer a general question about operational excellence all day without telling you anything. If you already work there and want a skip-level conversation instead, the questions for your own COO are the better starting point.

20 questions · each with a note on why · conversation guide

The questions

Open any question for the note

  1. Which three numbers would tell an outsider whether this operation is healthy?

    Why ask it

    Forcing a short list reveals what the COO believes the operation is for. If all three are activity counts rather than outcomes, you have learned something before you open a single report.

  2. Which operating metrics do you watch daily, and which can wait for the weekly review?

    Why ask it

    Separating heartbeat signals from review material shows whether the reporting cadence matches how fast the work moves. A COO who watches everything daily usually has a team that spends its mornings assembling numbers instead of using them.

  3. Where do you think our biggest throughput constraint sits right now?

    Why ask it

    The answer tells you whether they think in terms of a system with one binding constraint or a list of teams that all need to try harder. Ask what the queue looks like in front of it, since a constraint with no visible backlog is probably not the real one.

  4. How do you set capacity plans against the demand forecast?

    Why ask it

    This surfaces how staffing, shifts, and supply get matched to volume, and how much slack is deliberately held. Watch for a plan built on an average, because averages hide the peaks that actually break service levels.

  5. How do you measure quality alongside speed?

    Why ask it

    Every throughput push erodes something, and the interesting part is which measure they trust to catch it: rework rate, escapes, complaint volume. If quality is only measured by customers complaining, the cost has already been paid.

  6. Which processes here are ready to be simplified or retired?

    Why ask it

    A COO with a specific answer has been looking at cycle times rather than org charts. Naming a process nobody wants to defend is also the cheapest place for a new person to make progress in a first quarter.

  7. What manual work still has no clear owner?

    Why ask it

    Unowned manual work is where capacity quietly disappears, usually into a spreadsheet maintained by one person who is now unable to take a holiday. The answer often identifies both an automation target and a single point of failure.

  8. Where do we deliberately accept process or technical debt, and what is the plan to pay it down?

    Why ask it

    Accepting debt on purpose is normal; the tell is whether there is a date attached. Debt with no paydown plan is not a trade-off, it is an unpriced liability that will surface as an incident.

  9. How do you decide whether to build a tool or buy one?

    Why ask it

    You are looking for the actual test they apply, such as whether the process is a differentiator or just table stakes. A COO who always builds is paying maintenance forever; one who always buys ends up gluing eleven vendors together.

  10. Which vendor relationship worries you most right now?

    Why ask it

    This puts the single points of failure outside the company on the table: sole-source suppliers, one carrier, one payments provider. Ask what the fallback is and whether it has ever been tested, not just documented.

  11. How do you stop functions from optimising against each other?

    Why ask it

    Sales booking what operations cannot deliver is the classic version. Listen for a shared metric or a forum where the trade is made explicitly, because goodwill between department heads is not a mechanism.

  12. What actually happens here after something breaks?

    Why ask it

    The shape of the postmortem tells you the culture: a written timeline and a change to the system, or a meeting that ends with a person's name. If actions from the last one were never closed, the ritual is theatre.

  13. What is the escalation path when two teams block each other?

    Why ask it

    Without a named path, cross-team blockers sit until someone senior notices by accident. Ask how long a blocker is allowed to live before it must be raised, since a policy with no clock does not change behaviour.

  14. How do you replan when a quarter slips halfway through?

    Why ask it

    This tests whether replanning is a discipline or a panic. Good answers describe what gets cut and who is told; weak ones describe working harder, which is how the next quarter gets damaged as well.

  15. Which change to the org over the past year did the most for execution?

    Why ask it

    Attributing a result to a specific design choice, such as merging two handoffs or moving a team under one owner, shows they think about structure as a lever. Vague answers about alignment usually mean the reorg was about headcount.

  16. How deep is the bench if you lost two operations managers next month?

    Why ask it

    Key-person risk is the thing an outside reviewer misses most often, because it never appears on a process map. Push for names and cover plans rather than a general answer about succession.

  17. If you had to take ten percent out of the cost base, where would it come from first?

    Why ask it

    A COO who cannot answer quickly has not modelled it, which is itself the finding. The order they name things in tells you what they consider load-bearing.

  18. What would you spend the next headcount on, and what does that person fix?

    Why ask it

    The next hire is the clearest statement of where an operation is short. More of the same role means the constraint is volume; a new capability means the constraint is design.

  19. If you handed this operation to someone tomorrow, what would you warn them about?

    Why ask it

    Framed as a handover rather than a confession, so it gets the caveat that a direct question about weaknesses does not. Expect one system, one relationship and one deadline.

  20. What data do you wish you had and do not?

    Why ask it

    Instrumentation gaps are where decisions are currently being made on instinct. The gap they name is usually the highest-value thing you could go and build, and it is also a fair test of whether they know their own blind spots.

Talking to a COO

Practical guidance for the conversation itself

Prepare with their numbers, not yours

Get the weekly pack first

Ask for the operating review deck or dashboard before the meeting and read it. A COO will spend the whole slot narrating those numbers if you let them, and that time is better used on the one figure you could not explain.

Bring a specific object

Name a queue, a shift, a handoff, or a single order that went wrong. Operations people answer concrete questions concretely and general questions with philosophy, so the specificity of your question sets the quality of the answer.

If you are bringing them a problem

  1. 1State what is happening and how often, with a number attached rather than an anecdote.
  2. 2Say what it costs in hours, money, or missed commitments, so it can be ranked against everything else on their list.
  3. 3Offer the two options you would choose between, and say which one you would pick and why.
  4. 4Name what you need from them: a decision, a person, budget, or permission to stop doing something else.

If this is an interview

  • Ask what broke last quarter and what changed as a result. The answer describes the real learning loop better than any statement about culture.
  • Ask which number you would be judged on, who else influences it, and who currently owns it.
  • Ask how many layers sit between you and them, and how often they see work at your level.
  • Ask what happened to the last two people in the role you are discussing.

What a weak answer sounds like

  • Data driven as a self-description, with no metric named and no threshold for acting on it.
  • A postmortem process whose output is a person being more careful next time.
  • Known process debt with no owner and no date.
  • A capacity plan built on an average, with no view of the peak.
  • Every constraint described as a resourcing problem, which usually means the flow has never been mapped.