How do you handle disagreements with engineering?
This is a collaboration question. Not a story about who won. The interviewer wants to see that you respect engineering's judgement, bring data instead of authority, and commit once a decision is made.
Step 1. Structure (STAR, weighted to Action)
- Situation: what you disagreed about, in one or two lines.
- Task: the decision that had to be made, and why it mattered.
- Action: how you understood their view, what data you brought, how you found a path.
- Result: the outcome, a number, and the state of the relationship after.
Step 2. Know which kind of disagreement it is
- How to build it: engineering owns this. You bring constraints (time, users, cost), not the design.
- What to build and why: you own this, but you earn agreement with evidence.
- Estimates and deadlines: negotiate scope, not the estimate.
- Tech debt vs features: the hardest one. Both sides are right. It needs a trade-off, not a winner.
Step 3. What a strong answer shows
- You listened first, and can explain their position fairly.
- You used data and user impact, not "the business wants it".
- You found an option better than both starting positions, or you disagreed and committed.
- The relationship got stronger.
Step 4. Sample answer (adapt it to your own story)
"I was the PM for search on a fashion e-commerce app. Before the festive sale, I wanted to ship new filters. Our engineering lead wanted the quarter to rebuild the search service.
I asked him to walk me through his reasons. Search latency at p95 was 1.8 seconds, and we had three incidents during the last sale. He was right. A sale on slow search would cost more than missing filters.
I brought my data too. 30% of search sessions used filters, and the biggest gap was a size filter, which drove most exits. We agreed on a split. The team fixed the two slowest query paths first, which took three weeks. Then we shipped only the size filter on the existing service. The full rebuild went first on next quarter's roadmap, and I defended that slot in planning.
During the sale, p95 latency was 900 ms, with no incidents, and search conversion rose 6% (illustrative). The lead later asked me to co-present the rebuild plan to leadership."
Step 5. Mistakes to avoid
- A story where engineering was simply wrong.
- Escalating to your manager as the first move.
- Overruling engineering on how to build something.
- No result, or a result with no number.
Interview tip: say what you learned from the engineer. It shows the disagreement made the product better, not just that you got your way.
Community Answers (0)
No answers yet. Be the first to share your approach!