Tell me about a time you disagreed with someone senior

Asked in: behavioral round at every senior level.

The L4 candidate’s story ends with “and they realized I was right.” The L6 candidate’s story ends with a synthesis neither party had at the start, or with the candidate updating their position based on new information.

What the question actually tests

Three specific things:

Did you have the technical conviction to disagree at all? (Some candidates won’t, on principle, with anyone senior. That’s a signal.)
Did you raise the disagreement productively? (Email tantrum, calm 1:1, or formal escalation each say something different.)
Did you remain open to being wrong? (The most senior signal is updating your view when warranted.)

What an L4 answer sounds like

“My manager wanted to use approach A. I wanted to use approach B because I thought it was better. I argued for B and we eventually used B. It worked out.”

This describes a victory, not a disagreement. you don’t actually engage with the senior person’s reasoning, you just push.

What an L5 answer sounds like

A good answer follows a STAR(L) structure with specific technical content:

Situation: brief context, what the team was deciding.
Task: what you were responsible for in the decision.
Action: specifically what you did to surface the disagreement (1:1, written doc, larger forum), what evidence you brought, how you engaged with the other person’s argument.
Result: what was decided, what shipped, how it worked.
Learning: what you took away about disagreement, the technical area, or the people involved.

The story should:

Have specific technical content the interviewer can probe.
Show that you understood the senior person’s reasoning, not dismissed it.
Include some evidence-gathering or experimentation to inform the decision.
End with an outcome (and ideally a learning), not just “I won.”

What an L6 answer adds

The L6 version often has these properties:

The disagreement was about strategy, not just technique. “We disagreed on whether to invest 6 months in a custom solution or use an off-the-shelf tool” is an L6 disagreement; “we disagreed on whether to use Adam or SGD” is an L5 disagreement.
You facilitated the resolution. Brought the right people into the room, framed the trade-offs, ran a small experiment to inform the decision, wrote the proposal that the team eventually agreed to.
You changed your view at some point. Maybe started with strong conviction for B, learned something through the process, updated to a third option C that was better than either A or B.
The story includes someone you respect being on the other side. L6 candidates have learned to disagree well with smart people, not just to dismiss them.

Tells that get you a strong-hire vote

The story is specific and technical, not abstract.
You understood the other side’s reasoning before disagreeing.
You used evidence or experimentation to inform the decision.
You updated your view at some point in the process.
The outcome is a synthesis, not a clean win.

Tells that get you down-leveled

The story is generic (“I disagreed with my manager about a model”).
The other person is presented as wrong and you as right, with no grey area.
No evidence of evidence-gathering.
“I won” framing.
You can’t describe the other person’s argument convincingly.

A trap to avoid

Some candidates pick a disagreement where the senior person’s view was clearly wrong (a manager who didn’t understand ML pushing for naive approaches). These stories make you sound right but make the senior person sound dumb, which signals you don’t have actual senior peers to disagree with.

Better: a disagreement with a respected senior person whose argument was also defensible. Demonstrates you can engage with smart opposition.

Common follow-up

“What if you’d lost that argument?”

The L6 answer:

“Two layers. First, I’d own the chosen direction publicly; once a team decides, you commit. Second, I’d track the metrics that distinguish the two approaches and revisit the decision when there’s evidence to revisit on. If the chosen direction is failing on the metrics, that’s information; if it’s succeeding, that’s also information, including about my own judgment. The worst pattern is to hold a ‘I would have told you so’ grudge in case the decision goes wrong; that poisons the team and makes you less able to disagree well next time.”