The average chief data officer lasts about thirty months. CEOs last nearly seven years, and CFOs and CIOs around four and a half. MIT Sloan put those numbers side by side in early 2026, and the gap has nothing to do with chief data officers being worse executives. The cause is a seat with boundaries nobody agreed before the appointment.
The Chief Quantum Officer is younger than that role and shares its shape. So few people have the title that no tenure data exists yet, and by the time it does the first cohort will already have learned these lessons expensively. So I draw here on the failures that are visible now, since no dataset exists.
Five recur, and each has an early warning sign you can see months before the failure. You can recover from four of the five if you catch them early, and from none of them at the end.
The demo that became a roadmap
A vendor runs a proof of concept, and it works in the narrow sense of producing output on a real problem instance. Somebody senior sees it and asks what it would take to do this properly, and each manager who passes the answer upward makes it sound more certain. Eighteen months later the CQO is defending a delivery date nobody ever committed to.
The cause is compression rather than dishonesty: people shorten technical results each time they report them upward. Each management layer drops one of the three assumptions that qualified the lab result, and by the fourth retelling none remain.
Guarding against this is a communication discipline rather than a technical one. Whenever the quantum team reports a result, it should state the conditions in the same sentence, phrased so a non-specialist repeating it passes them on. Not “we ran a portfolio optimization on quantum hardware” but “we ran a scaled-down instance that classical solvers still handle faster, to learn how the tooling behaves.”
Early warning. Somebody senior repeats a result back and it sounds better than the original. Correct it at that moment, even though you will sound pedantic and unwelcome. The cost of correcting rises every month you wait.
The permanent pilot lab
The function becomes a place where interesting things happen and nothing ships. Each year the CQO adds a new pilot, a new partner, a conference presentation. Every year the answer to what changed in the business is that the team learned a lot. The team has learned a lot, but the business has gained little from it.
This failure feels comfortable, and that is the danger. Pilots are enjoyable and always yield a demo. Nobody has to say no to anyone. Nobody sees the failure until a new chief financial officer asks what the company got for the last four years of spending.
The fix is a structural one. For every funded pilot, write down the decision it should inform and its date, and make sure the organization would actually act on the outcome. If no result of a pilot would change anything, it is a training exercise. Training exercises are fine as long as you budget and label them as such rather than presenting them as progress.
Early warning. Two consecutive planning cycles where no pilot has been stopped. A team with judgment kills pilots, and a team that only adds them has quietly stopped thinking about which ones matter.
Capture
The CQO comes to see the field through one vendor’s frame. The CQO adopts the vendor’s roadmap and benchmark as their own, and stops asking questions that would embarrass the vendor.
This rarely looks like corruption. It looks like a good working relationship. The vendor’s team is responsive, technically excellent, and generous with their time, and after two years the CQO depends on them intellectually. In the interview piece I treat a single-vendor background as a hiring signal, but capture is not really about where someone came from. Independent people get captured in the seat too, through access rather than money.
You can prevent it with three habits. Keep an active relationship with at least one competing platform, even with no immediate use for it. Once a year, have somebody outside the function write the case against the current primary vendor. And notice which questions have become awkward to ask, since awkwardness is the tell.
Early warning. The last three technical briefings the executive team received all came from the same company. Nobody decided that, and no one noticed.
The sponsor who left
The CQO was appointed by a chief executive or a board member who understood why. Then that person leaves. The successor inherits an executive they did not appoint, running a function they cannot yet evaluate, on a budget line that is unusually easy to question.
People blame this one on bad luck more often than the other four, but it is the clearest failure of preparation. A mandate that depends on one relationship is a favour with a job title.
Preventing it is unglamorous work, and it takes about eighteen months. Build a second and third executive who benefit visibly from the function, usually the CISO and one business unit leader with a real decision in front of them. Get the mandate written into something durable, a board resolution or a charter or a risk appetite statement, so that it remains in force after the person who granted it leaves. Make sure somebody else’s work depends on at least one thing the function produces. People keep what they depend on through a reorganization, and goodwill alone is not enough.
Early warning. Ask who would defend this function’s budget if the sponsor were gone next month. If the answer is nobody, the failure has already happened and only the timing is undecided. If you count that number quarterly, the question becomes a standing instrument.
When the work is done but the role continues
The rarest failure and the one I find most interesting, since it looks like success right up until it does not.
The team completes the migration, and the CTO’s organization can now run the technology as a routine. The strategic questions that needed an executive have been answered, and what remains is operational. And the CQO keeps producing quarterly updates about a topic that no longer needs an officer.
I would rather see a Chief Quantum Officer wind the seat down deliberately than watch the board quietly stop listening; the difference between those endings is who decides. I will give this one its own piece.
Early warning. The quarterly board conversation stops changing at all. If the board asks the same questions in year four as in year three, either the reporting has gone stale or the work is done, and somebody should find out which.
What the five have in common
All five are failures of organizational judgment.
No Chief Quantum Officer loses the seat for misunderstanding surface codes or backing the wrong qubit modality. They lose it when a result gets repeated without its conditions, or a portfolio of pilots never becomes a decision, or a vendor relationship becomes a worldview, or a mandate depends on one person’s memory of why it existed.
It contradicts how companies make these hires. Companies searching for this role weight technical depth heavily and organizational judgment lightly, and then almost every CQO who fails does so on the second. In the MIT Sloan piece on chief data officers, Ajai Sehgal of the Mayo Clinic describes his own role in the same terms. Creating the seat means carving responsibility out of somebody else’s, and if the company gets that agreement wrong, the new officer has an adversary for the remainder of a tenure.
The quantum version of that sentence has not been written yet, since the tenures are not over. When it is, it will read much the same.