Questions to Ask Director of Engineering
Questions for a conversation with a director of engineering, whether you are a candidate deciding whether to join their organisation or on a panel hiring for the role. Covers org structure, how work is decided, delivery speed, on-call, promotions, and attrition.
The questions
Open any question for the note
How many engineers report up to you, and how many layers are between you and them?
Why ask it
This sets the scale of everything else they say. A director over four teams still knows individual engineers; a director over forty has to work through managers, and their answers about culture should be read as second-hand.
What has changed in the structure over the last year?
Why ask it
Reorganisations are where engineering time quietly goes. Ask how many there have been and what problem each was meant to solve, because two in a year usually means the first one did not work.
What are you personally spending your time on this quarter?
Why ask it
Compare the answer to how they described the job a minute earlier. A director who spends the quarter on hiring, or on one struggling team, is doing a different job from one working on architecture or planning, and only one of those matches what you were told.
How does work reach the teams, and who decides what gets built?
Why ask it
The interesting case is a disagreement between engineering and product. Ask who breaks the tie in practice, since an answer about collaboration with no named decision-maker usually means the loudest stakeholder wins.
How long does a change take to go from an engineer's laptop to production?
Why ask it
One of the few answers in an engineering conversation that is hard to spin. Minutes, days, and a monthly release window are three different working lives, and the reason it is slow tells you what has been tolerated.
What is on-call like, and when did you last carry it yourself?
Why ask it
Ask for the number of nights someone was woken in the last rotation. A director who cannot approximate it, or who has not been near the rotation in years, is unlikely to be the one who fixes it.
How does maintenance and technical debt get funded?
Why ask it
Look for a mechanism: a standing percentage, a reserved sprint, an owned backlog. If it depends on engineers arguing for it case by case, the answer is that it does not happen unless something breaks.
What happened after your last serious outage?
Why ask it
You are listening for whether the review produced changes anyone can name, and whether an individual was blamed. Organisations that cannot recall a recent outage are usually not monitoring closely enough to have noticed.
How do you know whether the organisation is doing well?
Why ask it
The metrics they reach for say what they optimise. Delivery counts, incident rates, retention, and internal survey scores each pull behaviour in a different direction, and an answer built only on velocity is a warning about how teams will be pressured.
How do promotions work here, and how many people were promoted last cycle?
Why ask it
A named process plus a number is a working system. A described philosophy with no recent examples means promotion depends on your manager's ability to campaign, which is worth knowing before you join.
How do you handle a manager who is not performing?
Why ask it
This is where a director's real willingness shows. Ask what happened the last time, and how long it took, since a weak manager left in place for a year is the most common reason good engineers leave an otherwise healthy company.
What do you do when a senior engineer and a product manager cannot agree?
Why ask it
Ask for a specific case rather than a policy. Whether they escalate, decide themselves, or leave the two to sort it out tells you how much air cover the engineers under them actually have.
What is attrition like, and where did the people who left go?
Why ask it
Both halves matter. People leaving for bigger jobs elsewhere and people leaving for the same job at a competitor mean different things, and a director who has not thought about the destinations has not asked.
How much of the hiring do you do yourself?
Why ask it
Directors who still interview know what the market looks like and what their bar actually is. Full delegation is defensible at scale but it means their descriptions of team quality are inherited from other people.
What is the biggest thing you are unhappy with in the organisation right now?
Why ask it
The most useful question in the conversation, because a director who names something specific, with a plan and a timeline, is showing you both candour and command. A polished answer with no real complaint means you are still in the sales conversation.
What decision here did you get wrong?
Why ask it
Look for a real cost and a real correction: a platform investment abandoned, a reorganisation reversed, a hire kept too long. An answer about learning to communicate more is a non-answer.
How do you keep enough technical context to be useful, and where do you rely on others?
Why ask it
Both extremes cause problems. Directors who still want to make technical decisions block their staff engineers; directors with no context cannot tell when an estimate is nonsense. Listen for how they describe the boundary they have drawn.
How is engineering time split between new work, maintenance, and unplanned interruptions?
Why ask it
Ask for approximate proportions rather than a philosophy. A director who has never measured this cannot tell you why roadmaps keep slipping, and the answer is usually in the third category.
A year from now, what would make you say the year went well?
Why ask it
This tells you what they are actually accountable for, which is often narrower than the vision they opened with. Notice whether the answer involves shipping something, changing the organisation, or a number someone above them cares about.
What would make you leave this job?
Why ask it
A blunt question that is fair at this level and usually answered honestly. If the director is likely to be gone within a year, everything they have promised about direction and structure is contingent, which changes how you weigh the whole conversation.
Talking to an engineering director
Practical guidance for the conversation itself
If you are the candidate
Spend your questions on mechanisms, not values
Every engineering leader will say they value quality, autonomy, and growth. Ask how technical debt is funded, how promotions were decided last cycle, and how long deployment takes, since those answers cannot be given aspirationally.
Check their answers against an engineer on the team
Ask to speak to someone two levels down and put two of the same questions to them: on-call load and how work arrives. Where the two accounts diverge is the actual state of the organisation.
Ask about the last time, not the policy
"How do you handle X?" invites the intended process. "What happened the last time X came up?" gets you what actually occurs, and the difference between the two answers is the useful part.
If you are hiring for the role
Make them do arithmetic out loud
Ask for headcount, attrition, promotion counts, and time-to-production as numbers. Candidates who have run an organisation carry these; candidates who have narrated one do not.
Probe the manager-management
The distinguishing skill at this level is managing managers, not engineers. Ask how they coached a struggling manager, how long they gave it, and what they did when it did not improve.
Test the technical boundary deliberately
Give them a scenario where a staff engineer wants to rewrite a system and the roadmap cannot absorb it. You are looking for someone who neither overrules on instinct nor abdicates the decision.
Answers to treat as warnings
- No number available for attrition, promotions, or deployment frequency.
- Technical debt handled entirely by engineers persuading their manager case by case.
- A recent outage with no changes anyone can name, or one attributed to a single person.
- Two or more reorganisations in a year, each described as fixing the last.
- No complaint about the organisation, from someone who has been there over a year.
- Success measured only by delivery volume or velocity.
- A promotion philosophy with no examples from the last two cycles.