The STAR Method Is Fine. Your Stories Are the Problem.

Situation, Task, Action, Result. Everyone knows it, most people can recite it, and it still goes badly — because STAR tells you how to shape an answer and says nothing about having one. The failure in a behavioural loop is almost never structure. It is running out of material in round three.

The arithmetic nobody does

A four-round onsite asks roughly three behavioural questions per round. That is twelve prompts. Candidates typically arrive with three solid stories, reuse each four times, and by the last round are visibly stretching one anecdote to cover leadership, conflict, failure, and ambiguity.

Interviewers compare notes afterwards. Hearing the same project four times from four angles reads as a thin career, whether or not that is true.

Build an inventory, not a script

Go back over the last three years and list every project, incident, disagreement, migration, launch and postmortem you were meaningfully part of. Aim for fifteen entries. You are not writing them up yet — just naming them, so they are retrievable under pressure.

Then map each to the dimensions it can evidence. Most entries cover two or three: a bad launch is a failure story, an ownership story, and often a conflict story depending on which part you tell.

The dimensions worth covering

• A failure you caused, with what changed afterwards.

• A disagreement with someone more senior.

• Something ambiguous you scoped yourself.

• A time you changed your mind given evidence.

• Work you did that nobody asked for.

• A deadline you missed, and how you handled the telling.

On the Result

The R is where good stories die. "It went well" is not a result. A number is better, but an honest small number beats an impressive vague one — "cut the p99 from 1.2s to 400ms" lands, and "significantly improved performance" does not.

If you cannot say what changed because of you, the story is about a project you were near rather than work you did.