Tell me about a project that failed.

Structure it in four parts:

1. The situation, briefly. One or two sentences.
2. What went wrong, factually and without blaming individuals.
3. Your contribution to it. This is the part being assessed. "I assumed the stakeholder's definition of active customer matched ours, and I never confirmed it" is a real answer.
4. What changed as a result — a specific, durable practice, not a sentiment.

The fourth part is what separates a good answer from a merely honest one. "I now write the metric definitions into the requirements document and get them signed off before building" is a change. "I learned to communicate better" is not.

The failure modes on this question are predictable:

The humblebrag — "the project failed because I cared too much about quality". Interviewers recognise this instantly and it costs more than a real failure would.

Blaming others — even when accurate, it answers a question that was not asked.

No specific change — a failure with no resulting practice reads as a story rather than learning.

Pick a genuine failure with a bounded, non-catastrophic consequence, and be specific about your part in it.