Tell Me About a Time You Received Feedback and Changed Your Approach
The question: "Tell me about a time you received feedback and changed your approach as a result. How did you implement the changes?"
Also asked as: "Tell me about feedback that changed how you work, not just what you thought." · "What's a piece of feedback you acted on, and how do you know it stuck?" · "When did you last change your mind about how you operate?"
Who's asking: an Engineering Director or Senior Engineering Manager, in a behavioral round.
Executive Summary: The director spends ten seconds on what you heard and the rest on what you did with it: what changed on Monday, what it cost, how you know it worked, and how you'd notice if it slipped. What grows with level is what the change costs and who pays: an hour a week of your own time at SDE, a funded bet at Principal, where the feedback was about your judgment, not your output.
What the director is really asking
Every manager has given feedback that was agreed with warmly and changed nothing. So the director is answering one question of their own: "If I tell this person something in their first month, will anything be different in their sixth, without me checking?" They listen for four things:
- A Monday. The first concrete thing you did, dated and small. "I worked on it" has no Monday.
- A cost. Your time, the team's time, a date, a bet. A free change was a small one.
- A check. How you know it worked, and what the person who gave the feedback would say now.
- Judgment about dose. Feedback says "less of this". You decide how much less. Over-correcting and then finding the middle beats a clean swing, because it shows you were watching.
This isn't the question about the moment the criticism landed (01-handling-work-criticism), or about going to look for feedback (050/03). Here the feedback can have arrived any way at all; the director probes the doing.
It maps most closely to Amazon's "Earn Trust" (vocally self-critical, and following through), and loosely to "Learn and Be Curious"; both approximate. Three ladder dimensions carry the weight: people, ownership (the change and its cost are owned to done), and how impact is measured.
Level fit at a glance
| SDE | Senior SDE | Staff | Principal | |
|---|---|---|---|---|
| Asked at this level? | Core | Common | Common | Common |
| Story scope they expect | A habit in your own work | How you run your team; a partner paid for the old way | How several teams experience the way you drive change | How an exec experiences your judgment |
| The bar in one line | A dated habit, the right dose, and the person who told you can see it | A team practice changed, costed honestly, with a date you moved and owned | Managers you don't report to told you how you operate; the first fix failed; several teams changed something they own | An exec corrected how you operate; you got the point on the second try; a bet moved against your own plan |
| Instant down-level | "I took it on board and worked on it" | "I started communicating more with the partner team" | "I wrote the standard and the director signed it" | "The exec asked for X, so I delivered X" |
SDE: find the dose, then show the habit
What meets the bar
- You can date the first change, and it was small.
- You over- or under-shot and noticed, from what it did to the people around you.
- It cost you something you can name, usually time, and there's a number and a check-back with the person who told you.
- Check your altitude: if your whole team changed how it works, it reads as Senior. If the story ends at "I agreed", it reads below the bar.
Model answer
"I was about two years in at a comics-reading app, building offline chapter downloads. In a one-on-one my tech lead said, 'You ship fast, but you ship alone. Twice this quarter I've found one of your decisions in a pull request after someone else had to work around it.' The example was storage. I'd saved each downloaded chapter as one encrypted file; the iOS reader streams panel by panel, so the iOS engineer spent four days writing a splitter he shouldn't have needed.
On Monday I started posting a short note in the team channel before I began anything: what I was deciding, and the option I'd rejected. Then I'd wait for one reply. Two weeks later I'd posted fourteen, including one about a log level. My lead said, 'Now I'm reviewing your thinking twice,' and a teammate admitted he'd muted the thread. I'd turned 'don't ship alone' into 'ask about everything', which was a different way of not using judgment.
So I wrote a rule for what gets a note: anything that changes a storage format, anything another client reads, and anything that's hard to undo. Everything else, I just build. That's about one note a week, and each costs me about an hour.
Before the feedback, two of my last six features had needed rework after review, about four days each, and not my four days. Over the next two quarters, none of eleven did.
At my next review the lead's comment had moved to 'the notes are how I know what you're working on.' I'd know it had slipped if a format change of mine ever shipped without one. When I take on feedback now, I write down what too far would look like before I start, so I notice when I get there."
Why this lands at SDE
- "anything that changes a storage format, anything another client reads, and anything that's hard to undo": the line only an SDE would say. A rule for their own work, no team practice, no authority.
- "I'd turned 'don't ship alone' into 'ask about everything'": they over-corrected, noticed from other people's reactions, and found the dose.
- "not my four days" and "the notes are how I know what you're working on": the cost landed on someone else, and the person who gave the feedback confirms the change.
Follow-ups to expect
- "You went from zero notes to fourteen in two weeks. What does that tell you about how you take feedback?": the most dangerous one here → it's asking whether you swing. Say so, then what you do before changing a habit now.
- "If I asked your tech lead whether that's still true, what would they say?": tests whether the change is real → the review comment, and your last note.
- "What did you learn that you still use?": the SDE probe → the "too far" line, and when you last used it.
Red flags
- ❌ "I took it on board and became more collaborative." No Monday, no cost, no number.
- ❌ "I now check every decision with my lead." Judgment swapped for permission; no dose found.
Senior SDE: change the team's practice, and own what it costs the roadmap
What meets the bar
- The feedback is about how you run your team or project, not about your code.
- You changed a team practice the team had to carry, estimated its cost and missed, and told the people who'd pay before they found out.
- There's a system or product number with the cost beside it, and the partner who complained can see the difference.
- Check your altitude: your own habit reads as SDE. If managers of several teams changed what they own, it reads as Staff.
Model answer
"I led a team of six at a greeting-card printing company, owning the renderer that turns web designs into the plant's press-ready PDFs. At my mid-year review my manager read me two comments, from the plant's prepress lead and our PM: 'You run the team as if the plant were a downstream system. They're the customer, and they find out about your changes when the press is running.' That year we'd caused four press-stop reprints, about $31,000 plus night shifts; one font-fallback change had ruined 1,800 cards.
That Monday I proposed to the team that every renderer release gets a physical proof at the plant, signed off by the prepress lead. I estimated half a day per release. It cost about three. The plant could fit us into a live press queue twice a week at best, and every proof needed one of my engineers there for an afternoon with a loupe. Our foil-stamping layer slipped from mid-September to mid-October and missed the holiday catalog print, which the PM priced at about $70,000 in foil-card revenue for the season.
I told the PM and the prepress lead the date was moving a week before it was public, and I said the estimate was mine. Then I re-cut it: physical proofs only for changes to fonts, color profiles, bleed or trim. Everything else gets a pixel diff against the previous renderer's output, which two of my engineers took three weeks to build. I kept the physical step because the plant's sign-off was the point of the feedback, not the proof itself.
Over the next year we had no press-stop reprints from renderer changes, and one CMYK profile change was caught at proof that the diff wouldn't have seen. The proof loop still costs about half a day a week, and the prepress lead sits in our release planning every two weeks. My manager's next review quoted her: 'I've stopped keeping a watch list of their changes.' The day she stops turning up to release planning is the day I start worrying again. When feedback changes how my team works now, I estimate it with the people whose time it costs before I commit a date."
Why this lands at Senior SDE
- "I proposed to the team that every renderer release gets a physical proof at the plant": the line only a Senior would say. A practice on the team they lead, one partner outside it, no other team's manager involved.
- "I estimated half a day per release. It cost about three.": the miss as a number, and "I said the estimate was mine" owns it.
- "missed the holiday catalog print, about $70,000" and "I kept the physical step because the plant's sign-off was the point": the result is priced net, and the re-cut keeps the feedback.
Follow-ups to expect
- "Your PM lost the foil feature for the season. Did they agree with that trade?": prepare this one first → what the PM actually said, and that you'd have cut earlier if you'd estimated with the plant. Don't make the plant the reason.
- "You estimated half a day and it cost three. What did that tell you?": tests self-awareness → you didn't know their schedule, which was the feedback in another form.
- "How did you get the team aligned on a step that ate their week?": the Senior probe → who pushed back, and what changed because of them.
Red flags
- ❌ "I started joining the plant's weekly call." Your calendar, not your team's practice.
Quick check before moving on:
Staff: when a memo doesn't move five teams, change who carries it
What meets the bar
- The feedback came from managers you don't report to, about how your cross-team work lands on them.
- You changed how you drive change across teams, the first attempt failed, and several teams' managers changed something they own at a stated cost.
- You can say why it was yours and what you stopped doing to make room.
- Check your altitude: one team's process and a partner is Senior. If an exec changed what the business funds, it's Principal.
Model answer
"I was a staff engineer on the platform for a cinema chain: 300 screens, five product teams. After a reorg the shared venue-config service had no owning team, and I'd become its reviewer by default. In a 360, three of the five team managers said, in different words, 'You find problems for other people to own. You tell us what's wrong and leave.'
The same month, a Friday 6pm concessions release changed the venue-config schema and ticket scanners went down for 50 minutes at twelve cinemas, about $30,000 in refunds, and the sixth peak-window incident from a shared change that year. I committed to carrying anything I raised across teams to done. Then I did the thing I'd just been told about, in better clothes: I wrote a peak-window release freeze, got my director's signature and sent it to the five managers.
Four weeks later two teams had adopted it, two hadn't, and there'd been a smaller repeat. The loyalty team's manager told me, 'You sent us a rule. Who's there at 3pm on a Thursday when my on-call has to decide whether a fix counts as a shared change?' I asked each manager to name a release captain, and proposed the five captains take over the concessions team's release-planning meeting. The captains wrote the definition of a shared change and the exceptions process, in their calendar tool, not my document.
Four hours a week for a quarter, and I handed a payments-migration design I'd been leading to a senior engineer on the payments team. About 15 percent of releases moved to Monday through Wednesday, which costs the concessions team about two days a sprint. Over the next year, peak-window incidents from shared changes went from six to one, and refunds from about $110,000 to about $9,000. The captains extended the freeze to Christmas week without asking me. In the next 360 the same manager wrote, 'He turns up with the problem now and stays until it's ours.' Before I change anything across teams now, I ask who has to carry it on a Thursday afternoon, and I go and work with them."
Why this lands at Staff
- "I asked each manager to name a release captain": the Staff beat. Managers who don't report to them give up something they own, for a rule their own people wrote.
- "Then I did the thing I'd just been told about, in better clothes": the first fix failed for the reason the feedback named, and they say so without flinching.
- "had no owning team" answers "why was this yours?" before it's asked; "about two days a sprint" prices the result in another team's time.
Follow-ups to expect
- "You were told you drop problems on teams, and your first move was a memo. Why should I believe the second version stuck?": the most dangerous one here → what the captains did without you, and what you check.
- "Why was this yours? Five managers could have fixed their own release process.": the Staff probe → the unowned service and the incident count.
- "How did you get the other managers to agree to lose Thursdays?": tests influence without authority → the $110,000, and their captains set the rule.
Red flags
- ❌ "I wrote the policy and the director approved it, so teams had to follow it." The failed first version, told as the success.
- ❌ "The teams that ignored the memo caused the second incident." Blaming the audience means you didn't take the feedback.
Principal: get the letter, then go back for the point
What meets the bar
- The feedback came from an exec, about how you operate; your first change did what was asked and missed what was meant, and you went back to find out.
- A funded bet moved against your own plan, you can price what that cost, and the org kept a mechanism that has since stopped something of yours.
- Check your altitude: an exec in the room doesn't make it Principal. If no funded direction changed, it reads as Staff.
Model answer
"I set the technical direction for the replenishment and inventory platform at a convenience-store chain, about 3,000 stores. In a quarterly exec review, the COO said, 'You bring us decisions already made. By the time I see your roadmap, my only move is yes.'
I did what she asked, as I understood it. I set up a monthly options review for her staff: three options for every open decision, about fifteen a month, three weeks of my team's preparation a cycle. Two months in, she stopped me in the corridor. 'I asked to be in the decision, and you've given me homework.' She'd already told my CTO I'd answered a different question. I asked what she'd have done differently with the last year if she'd been in earlier. 'I'd have stopped you improving a forecast my store managers override a third of the time.'
Store managers were overriding 34 percent of automatic orders, and I'd treated that as a training problem. She meant store operations should help choose which problems the platform worked on. I cancelled the options review and proposed a quarterly direction review on her chief of staff's calendar: I bring the bets I'm considering dropping, and her ops directors bring what stores fight most.
At the first one, I took the forecasting-model upgrade off the roadmap, about $8 million a year in waste reduction, and put a store-override redesign in its place: managers adjust orders with a reason, and the reasons feed the model. The upgrade slipped fifteen months, about $10 million foregone by finance's math. A year on, overrides are at 18 percent, and out-of-stocks on the top 200 products are down 1.4 points, about $20 million a year in recovered sales by merchandising's number. The platform teams did the engineering; the bet was mine to move.
That forum stopped my shelf-sensor pilot at its third session, and I agreed. I'd know it had slipped if something shipped that no ops director had asked for. When an exec gives me feedback now, I write back one sentence saying what I think they mean, and ask them to correct it before I change anything."
Why this lands at Principal
- "I took the forecasting-model upgrade off the roadmap, about $8 million a year in waste reduction, and put a store-override redesign in its place": the line only a Principal would say. A funded direction moved against their own plan, in the exec's forum.
- "I did what she asked, as I understood it": the letter-not-point mistake, stated without excuse.
- "That forum stopped my shelf-sensor pilot at its third session, and I agreed" and "the bet was mine to move": the mechanism outlived the story and applies to them; "I" for the direction, the teams for the work.
Follow-ups to expect
- "Isn't deciding which problems the platform solves your job? Why hand it to the COO?": the most dangerous one here → you didn't hand it over. What you still decide, what she decides, and what the 34 percent said about who'd really been deciding before.
- "How did you bring the COO back after she said you'd answered a different question?": the Principal probe → the corridor question, and that she sets the review's agenda.
- "What mechanism outlived your involvement, and what wouldn't it catch?": the Principal probe → the quarterly review, who owns it, and that it can't see a problem stores haven't noticed yet.
Red flags
- ❌ "The COO asked for earlier visibility, so I built a monthly review, and she was happy." The first version, and directors have heard execs say "happy" when they mean "tired".
- ❌ "I pushed back, because roadmap decisions are the Principal's call." A Principal the COO can't correct is the most expensive kind.
Quick check before moving on:
One story, four altitudes
| SDE | Senior SDE | Staff | Principal | |
|---|---|---|---|---|
| Situation | Your lead: "you ship alone"; four days of someone else's rework | Plant and PM: "the plant is your customer"; four reprints | Three managers: "you find problems for others to own"; six peak incidents | The COO: "my only move is yes" |
| Your role | SDE on offline downloads | Lead of a six-person renderer team | Staff; default reviewer of a service nobody owned | Sets the platform's direction |
| The key move | A decision note; fourteen in two weeks; then a rule | A proof step at the plant; re-cut after the estimate missed | A memo that moved two teams; then captains who wrote the rule | Options review that missed the point; then a direction review she owns |
| Result | Rework 2 of 6 to 0 of 11; an hour a week | No press-stop reprints; foil feature four weeks late, about $70k | Peak incidents 6 to 1; refunds $110k to $9k; 15% of releases moved | Overrides 34% to 18%; about $20M/yr recovered; $10M upgrade deferred |
| Reflection | Write down what "too far" looks like first | Estimate the change with the people who'll pay for it | Ask who carries it at 3pm Thursday | Write back what you think they mean, and ask |
Draft your answer
Step 1: Pick the right story
Run each candidate story through the test for your target level:
- SDE: Did feedback about your own work become a habit you can date, and has the person who told you seen the change?
- Senior SDE: Did you change a practice your team carries, and did you own what it cost with the people who paid?
- Staff: Did managers you don't report to tell you how your cross-team work lands, and did at least two teams change something they own?
- Principal: Did an exec correct how you operate, did a funded bet move against your own plan, and did the org keep a mechanism?
Then check all of these:
- It's from the last three years, you were central, you can say what you did the next working day, and you can name the cost.
- Something went wrong in the implementation: an over-shoot, a missed estimate, a channel that moved nothing, or the letter without the point.
- You've checked with the person who gave the feedback. If you haven't, do it before the interview.
If you're a new grad. Feedback from a research advisor, an internship mentor or a project teammate counts if it was about how you work, not a grade, and the change outlasted the conversation. The director accepts a shorter run, one internship or one term, but still wants the first thing you did differently, what it cost and who saw it stick. "My advisor said my weekly updates listed what I'd tried, never what I'd learned."
Aiming a level above your current title? Tell your own level's story honestly and show you can see the next one. A Senior aiming at Staff tells the team-practice story and adds: "Next time, the people who'd carry it would write the rule; here I only had to persuade my own team." That beats a borrowed cross-team story that collapses at "how did you get the other managers to agree?".
Step 2: Fill the skeleton
SCOPE + FEEDBACK (20s) — Product and role; who said what, close to their words,
and the example they pointed at: ______
MONDAY (15s) — The first concrete thing you did, and how small: ______
WHAT WENT WRONG (20s) — Over-shoot, missed estimate, channel that moved nothing,
or the letter without the point: ______
THE RE-CUT (25s) — What you changed about the change, "I …, because …",
and what you kept: ______
LEVEL BEAT (inside THE RE-CUT) — the line only your level could say.
SDE: the rule for your own work. Senior: the team practice and the date it moved.
Staff: what other managers changed. Principal: the bet that moved, in whose forum.
RESULT (15s) — One number, with its cost beside it: ______
EVIDENCE IT HELD (10s) — What the person who told you says now, or something
the change did without you: ______
LANDING (5s) — The habit you kept from this, in one line: ______
Aim for about two minutes, with a 30-second cut ready (what they said, Monday, what went wrong, the number and its cost).
Step 3: Calibrate it
| If your answer says… | It reads as… | Fix |
|---|---|---|
| "I took it on board and worked on it" | No implementation | What you did the next working day, with a cost and a number |
| "I agreed right away and it worked" | Too clean to be true | What you tried first, and what went wrong with it |
| "I changed completely" | Over-correction, no judgment | What you kept, and how you found the middle |
| A Senior answer: "I started communicating more with the partner team" | SDE | A practice the team carries, and what it cost |
| A Staff answer: "I wrote the standard and the director approved it" | Senior or below | Whose hands carried it, and what each manager changed |
| A Principal answer: "The exec asked for X and I delivered X" | Compliance, not judgment | What they meant, and the bet that moved |
| "And it stuck" | Unverified | The check you ran, and what would tell you it had slipped |
Step 4: Rehearse the follow-ups
Prepare these five, and say each out loud once:
- "If I called the person who gave you that feedback, what would they say has changed?" The most dangerous one at any level. Ask them first.
- "What did it cost, and who paid?" A number, and someone who isn't you.
- "Your first attempt didn't work. Why should I believe the second one stuck?" Something the mechanism did without you, and what you check.
- "What did you decide not to change?" Feedback you set aside on purpose, and what that cost.
- "How do you take on feedback differently now?" One habit, and the last time you used it.
Common mistakes
- Spending the time on how you felt when you heard it. That's the criticism question. Here, get to Monday within thirty seconds.
- A clean swing. Real changes over-shoot, miss their estimate, or land in the wrong channel first. Say what went wrong and what you did about it.
- From Staff up, telling it as the managers or the exec changing. The feedback was about you. If your story ends with them adopting your rule, you've told it backwards.
- A reflection that undercuts the story. End on what you do differently now, not on how often you still get it wrong.
Key takeaways
- The question is about implementation: a dated first step, a cost, a check, and the right dose.
- As level rises, who pays for the change moves from you, to your team, to other managers, to the roadmap.
- Dose is judgment. Say what you kept, not only what you changed.
Test yourself on this question's key ideas: