In a STAR answer, when should a candidate say "I" and when is "we" the honest choice?
answer
- Who exactly is being scored here?
- Two pronouns, two different jobs
- Shared background versus owned decisions
- Delete every sentence lacking a first-person verb
basics
~20 sUse 'we' only for shared context — the team, the system, the goal that was genuinely collective — and 'I' for every decision, action and tradeoff you personally owned. An interviewer can only score what one named person did.
solid answer
~40 sThe two pronouns have two different jobs. 'We' sets the scene: the team owned the model retraining pipeline, the stale scores were a shared problem, the fix shipped as a team. 'I' carries everything the interviewer is actually scoring: what I looked at, what I decided, what I wrote, which tradeoff I accepted. So I keep 'we' in the two context sentences at the top and in the outcome, and I make sure every action beat has one first-person verb that is mine alone. The test I use is simple — if you deleted every sentence that does not contain 'I', would anything assessable be left? If the answer is no, the story is describing a team, not a candidate.
go deeper
Be ready to state the split in one line: 'we' for the shared setup, 'I' for every action you took. Rehearse one story out loud and count how many first-person verbs survive in it.
Expect to explain why the split exists — the scorecard carries one name — and to show the mechanics: context in the plural, one owned action per beat after that, and the outcome attributed honestly to the team plus your link to it.
Show judgment about where the credit boundary actually sits: which calls were yours, which were the team's decision you executed well, and how you say that difference aloud without hedging into vagueness or inflating.
Own the tradeoff explicitly: too much first person costs you the collaboration signal, too much plural costs you the scorecard, and the right mix differs per story and per listener. Decide it deliberately rather than defaulting to the register you use at work.
## The problem this rule solves A behavioral interviewer leaves the room and writes a scorecard about **one person**. Their notes have to say what *you* decided, built, escalated or measured. If your answer narrates a team win entirely in 'we', there is nothing individually assessable in it, and the interviewer has only two moves: ask the attribution follow-up, or score you conservatively on the evidence they have. Most candidates who lose points here are not lying and are not shy — they are describing the work exactly the way they would describe it in a team standup, where 'we' is the correct and collegial register. The interview is a different room. ## The split, beat by beat | Beat | Honest pronoun | Why | |---|---|---| | Situation | mostly 'we' | The team, the system and the problem really were shared. Two sentences is usually all this needs. | | Task | 'I' as soon as it narrows | 'We had to stop the bad scores' is context; 'I took the freshness path' is your task. | | Action | 'I', one owned action per beat | This is the scored part. Every beat gets one first-person verb. | | Result | 'we' for the outcome, 'I' for the link | The team shipped; you say which part of the shipped thing was yours. | The register that follows from this is: **'we' reserved for the two context sentences, one owned action per beat after that.** ## A worked pair A candidate is describing an incident on a machine-learning team that owned a fraud-scoring model and its retraining pipeline. *All-'we' version:* "We noticed the retraining job was quietly producing stale scores, so we rolled back to the previous snapshot and we added a check so it could not happen again." Everything in that sentence may be true, and none of it is scorable. A behavioral interviewer reading their own notes a day later cannot tell whether the candidate found the bug, ran the rollback, wrote the check, or sat in the channel while three other people did. *Re-attributed version:* "The team owned the retraining pipeline behind our fraud-scoring model, and stale scores had started reaching the decision service. [context] **I** pulled the metadata from the last three retraining runs and found the feature-freshness step exiting early without failing. **I** argued for rolling back to the previous training snapshot before we debugged, because the scores were live. **I** wrote the freshness assertion that made the job fail loudly instead of silently. We shipped the rollback together the same day, and the assertion is what has caught it since." Same story, same team, same honesty — and now the scorecard writes itself. ## The over-correction The opposite failure is real and it is more expensive at senior levels: an answer narrated entirely in 'I' for work that plainly took a team reads as credit-stealing, and interviewers do notice. The fix is not to sprinkle 'we' back into your action beats — that undoes the whole exercise. It is to keep the team **visible in the context and the result**, and to name specific colleagues' contributions where they were decisive: "the on-call engineer caught the alert; I traced it to the freshness step." Naming someone else's real contribution costs you nothing and makes your own claim more believable, not less. ## Hedged verbs are the same defect wearing a disguise 'I was involved in', 'I helped with', 'I supported', 'I was part of the effort that' — these are first-person on the surface and collective underneath. They are also precisely the phrases that trigger an attribution follow-up. Replace them with verbs that name a discrete act: decided, chose, wrote, measured, escalated, proposed, rejected, rewrote, paged, negotiated the scope of. ## How to check a story before you tell it Speak the answer once, transcribe or record it, and mark every 'we'. Re-decide each one as **we-for-context** (keep it) or **I-for-action** (rewrite it). Most first drafts of a team story lose the large majority of their 'we's in this pass, and the ones that survive are the two setup sentences and the shipping sentence — which is exactly where a listener expects a team to appear.
- If the work really was shared, isn't saying 'I' just taking the team's credit?No, as long as the team stays visible where it was real. The team appears in the setup and in the outcome; the first person appears on the acts that were yours — what you traced, argued for, wrote. Naming a colleague's decisive contribution out loud is part of the same discipline, and it makes your own claim more credible rather than less.
- Where in the Result beat does 'we' still belong?On the shipping and the shared outcome: the team rolled the change out, the alerts stopped. Then add one sentence linking your action to that outcome — the assertion you wrote is what catches it now. That pairing gives the interviewer both signals at once: a collaborator, and an individually assessable contribution.
- What is wrong with saying 'I helped with the rollback'?It is first person on the surface and collective underneath — it names no discrete act, so the listener still cannot tell what you did. It is also the exact phrasing that provokes the attribution follow-up. Say what the help consisted of: you ran the rollback, or you wrote the check, or you made the call to roll back before debugging.
saying these in an interview costs you the question
- Narrating an entire team win in 'we', leaving no assessable action
- Using 'we' for decisions the candidate personally made
- Hedging with 'I was involved in' or 'I helped with' instead of a real verb
- Over-correcting into all-'I' for work a whole team plainly did
- Treating the pronoun choice as modesty rather than as attribution