How do you close a STAR story honestly when the project failed or its outcome is still unresolved?
answer
- A negative ending is still an ending
- Do not soften, distribute, or pre-empt it
- Reflection carries the weight in failures
- For unfinished work, give the leading indicator
- Name the checkpoint that will settle it
basics
~20 sState the real ending plainly in past tense, including the negative one, then let the reflection sentence carry the weight. For unfinished work, give the leading indicator so far and the checkpoint that will settle it.
solid answer
~50 sA negative or unfinished outcome is still an outcome; the only unusable ending is a missing one. For a failure I say what happened without softening it — the migration was rolled back after two nights and the old path stayed in place for another quarter — and then spend the reflection sentence on what I changed as a result, naming the later occasion where it showed. That second sentence does more work here than in a success story, because it is the evidence that the failure was converted into something. For work still in flight I give the leading indicator I actually have and the checkpoint that will settle it: three of the eleven services are migrated and the failure rate on those has held at zero for 23 nights, with the rest scheduled behind the next freeze. What I do not do is describe the result the project would have had.
go deeper
Practise saying a negative ending in one plain past-tense sentence with no softening words, then one sentence on what you changed afterwards. Do not skip the ending because it is uncomfortable.
Explain why the reflection sentence carries more weight in a failure story than in a success one, and why context stated before the outcome reads as an excuse while the same context after it reads as analysis.
Demonstrate judgement about which failures are worth telling, how to report unfinished work as a leading indicator plus a checkpoint, and how to name a disconfirming condition you set in advance.
Own the call on how much of a bad outcome to surface unprompted, balancing candour against the impression it leaves, and be able to defend where you drew that line.
## Two different awkward endings They are usually treated as one problem and they are not. - **The failed outcome.** The work finished and it did not work. There is a result; it is negative. - **The pending outcome.** The work is real and unfinished, so the result does not exist yet — only early signal does. Both are commonly handled by simply omitting the close, which produces the worst version of each: a story that trails off, leaving the listener to assume the ending was bad *and* that you were hiding it. ## Closing a failure The structure is the same two sentences, with the weight shifted to the second. **Outcome sentence — plain, past tense, unhedged.** "We rolled the migration back after the second night, and the old path stayed in production for another quarter." Resist three instincts: softening it ("it was partially successful"), distributing it ("the team decided to roll back"), and pre-empting it ("to be fair, the requirements were unclear"). **Reflection sentence — where the value now lives.** "What I took from it was that I had no way to compare the two paths under real load, so I now put a shadow-read in front of any cutover — I did that on the next two migrations and caught a mismatch on the first one." That sentence is doing the job the metric would have done in a success story: it is the observable, checkable thing that came out of the work. A useful test: could someone who was there dispute either sentence? If your outcome sentence would make a former teammate wince at the spin, rewrite it. ### What not to add The conditional is the standard escape hatch here, and it is the failure mode this whole pitfall turns on. "If we had had another sprint I would have got the failure rate down" reads as a defence, and defences invite the follow-up you least want. Say what happened; the listener already knows some projects fail, and is watching what you do next rather than the outcome itself. ## Closing a pending outcome Here honesty means being precise about *where the evidence currently stands*, in three parts: 1. **The leading indicator you actually have.** "Three of the eleven services are on the new consumer, and those three have failed on none of the last 23 nights." 2. **The checkpoint that will settle it.** "The remaining eight go behind the next freeze, and the number we are watching is the nightly failure count across all of them." 3. **What you would call a failure.** Naming your own disconfirming condition is unusually strong signal, because it shows you set one in advance rather than planning to declare victory whatever happens. The register to avoid is the projection stated as a result — "it will bring the failure rate to zero" — which converts an unfinished story into an unverifiable claim, and does so in the exact sentence the listener will remember. ## Choosing whether to tell it at all Judgement, not rule. A failure story is worth leading with when the question asked for one, when the reflection is genuinely strong, and when the failure was in your control — a story about someone else's decision going wrong teaches the listener nothing about you. It is worth holding back when you have a clean-ending story that covers the same competency and the failure adds no distinct lesson. For pending work, the practical question is whether the leading indicator is strong enough to be interesting. Two weeks of clean nights across three of eleven services is interesting. "We have started" is not; if that is all you have, tell a finished story instead and keep this one as supporting material. ## Rehearsing the uncomfortable version This is where a mock-interview partner earns their time, with two instructions. First: interrupt any softening word — partially, sort of, more or less — the moment it appears in the outcome sentence. Second: after the close, ask "so what would you say the result was?" If your answer to that question is different from the sentence you just said, your close was doing spin rather than reporting, and the version you gave under the follow-up is the one that should have been in the story. Write these closes down like any other. A failure story that has not been written down is the one most likely to drift into the conditional under pressure, because the conditional is where the discomfort naturally pushes it.
- Should a candidate ever volunteer a failed outcome when the question did not ask for one?Yes, when the failure is the true ending of a story you have chosen for other reasons, and when the reflection is strong. Do not go hunting for failures to display humility, and do not pick one whose cause was entirely someone else's decision — the listener learns nothing about you from it.
- How do you avoid sounding defensive while giving context for a failure?Put the outcome first and the context after, never the reverse. "We rolled it back after two nights; the load profile in production was different from anything we had tested against" reads as analysis. The same two clauses in the opposite order read as an excuse arriving before the admission.
- What if the work is unfinished and the early signal is genuinely bad?Say so and say what you changed because of it: the signal that made you stop, the adjustment, and what you are watching now. An unfinished story with a bad early reading and a visible correction is stronger than one with no reading at all, because it shows you were measuring.
saying these in an interview costs you the question
- Trailing off rather than admitting the outcome was negative
- Describing the result the project would have had if continued
- Softening with partially successful or more or less worked
- Stating a projection about unfinished work as if already measured
- Blaming the outcome on someone else's decision and stopping there
- Giving context for the failure before stating the failure