{"id":947,"date":"2026-11-03T17:52:00","date_gmt":"2026-11-03T16:52:00","guid":{"rendered":"https:\/\/grindloop.ai\/blog\/?p=947"},"modified":"2026-11-03T12:26:06","modified_gmt":"2026-11-03T11:26:06","slug":"unpopular-decision-interview-answer","status":"publish","type":"post","link":"https:\/\/grindloop.ai\/blog\/unpopular-decision-interview-answer\/","title":{"rendered":"Why Your Unpopular Decision Interview Answer Sounds Like You Won the Argument"},"content":{"rendered":"<p>An unpopular decision interview answer fails when it sounds like you won an argument. Confidence is the easy part to show. What the interviewer can&#8217;t see yet is how you decided and who paid for it. They also want to know whether anything could have changed your mind. So build the answer in four parts. First, name the decision and why it was yours to make. Second, say what the people who disagreed were right about. Third, say what would have made you reverse it. That condition has to predate the change. Fourth, report the result against that condition, including anything you adjusted. The third part proves the decision had a test you ran. For an engineer, the decision might be a release rule, a library choice or a code review policy. The four parts stay the same.<\/p>\n<h2>An unpopular decision interview answer tests the decider&#8217;s half of &#8220;disagree and commit&#8221;<\/h2>\n<p>The question maps to one of Amazon&#8217;s leadership principles, <a href=\"https:\/\/www.amazon.jobs\/content\/en\/our-workplace\/leadership-principles\" target=\"_blank\" rel=\"noopener\">Have Backbone; Disagree and Commit<\/a>. Amazon&#8217;s page carries no date. It says leaders challenge decisions they disagree with. It then adds, &#8220;Once a decision is determined, they commit wholly.&#8221;<\/p>\n<p>Most candidates read that from the side of the person who disagrees. This question puts you on the other side. You made the call. Other people had to commit to it. Jeff Bezos covered that side in his <a href=\"https:\/\/www.aboutamazon.com\/news\/company-news\/2016-letter-to-shareholders\" target=\"_blank\" rel=\"noopener\">letter to shareholders for 2016<\/a>, sent in April 2017. He wrote, &#8220;If you&#8217;re the boss, you should do this too.&#8221; So expect the interviewer to listen for two things. Did you hear the objections before you decided? And did you leave the others a way to be proven right?<\/p>\n<p>Hello Interview, an interview-prep company, makes a related point in its <a href=\"https:\/\/www.hellointerview.com\/community\/questions\/unpopular-decision\/cmijj0g5c030i07ad7reu9ky1\" target=\"_blank\" rel=\"noopener\">guide to this question<\/a>. It warns against telling the story as standing firm and winning. It doesn&#8217;t cover reversibility, or what the critics get if they turn out to be right. The rest of this post does, with one example.<\/p>\n<h2>Tomas froze feature merges before every release<\/h2>\n<p>Tomas is a senior Android engineer on a team of six. His manager made him the release owner for a shopping app. The previous quarter had two hotfix releases. Both came from features merged in the last two days before a release cut. So Tomas decided on a merge freeze. For five working days before each release, only bug fixes could merge into the release branch.<\/p>\n<p>The decision was his to make, because release process was his remit. It was still unpopular. Two engineers had features planned for the next release. The product manager, Ines, had promised one of them to marketing. Jonas, the most senior of the two engineers, argued the freeze would just move the rush earlier. Tomas listened and decided anyway.<\/p>\n<h2>Say what the people who disagreed were right about<\/h2>\n<p>An answer that treats the other side as wrong sounds like a victory lap. So give their objection its strongest form. Then say which part of it you accepted.<\/p>\n<p>Jonas had a real point. A freeze can shift the crunch to the week before it. Ines had one too. A slipped feature costs a marketing date. That cost is real. Tomas said so in the planning meeting. He also changed one thing because of it. Features could still merge during the freeze behind a remote flag that stayed off. That answered Ines&#8217;s concern in part. Code could land early while the launch waited.<\/p>\n<p>Saying this in the interview shows you didn&#8217;t need everyone to be wrong.<\/p>\n<h2>Set what would reverse the decision before the change starts<\/h2>\n<p>Bezos also wrote in that letter that &#8220;Many decisions are reversible, two-way doors.&#8221; A merge freeze is one of those. So Tomas treated it as a trial with an exit.<\/p>\n<p>He wrote two conditions in the team&#8217;s planning doc before the first freeze. If crash-free sessions didn&#8217;t improve over three releases, the freeze would end. It would also end if more than one planned feature slipped per quarter because of it. He asked Jonas to run the review after the third release. Jonas was the strongest critic, so his review would carry weight with the team.<\/p>\n<p>The conditions gave the decision a test. They also gave the people who lost the argument a real way to win it later. A condition set in advance holds up under one follow-up. Ask yourself whether anyone else saw it before the change started. If no one did, don&#8217;t claim one. The last section covers that case.<\/p>\n<h2>Report the result against the conditions you set<\/h2>\n<p>Compare the result with what you said would count. Don&#8217;t just give the good number.<\/p>\n<p>Over three releases, crash-free sessions rose from 98.9% to 99.6%. There were no hotfixes. One feature slipped. That was inside the limit. Jonas&#8217;s review found something else. Two engineers had merged large changes on the last day before each freeze. So the crunch had moved, as he predicted. The team kept the freeze but cut it to three days. They also added a size limit on merges in the day before it.<\/p>\n<p>Keep that adjustment in the answer. The decision survived an honest review. The review still changed it.<\/p>\n<h2>Tomas&#8217;s answer, as he might say it<\/h2>\n<p>We wrote this answer for this post. It runs about ninety seconds aloud. Replace the release details with a decision you actually made.<\/p>\n<blockquote>\n<p>Last year I owned releases for our shopping app. We&#8217;d shipped two hotfixes in one quarter. Both came from features merged in the last two days before the release cut. So I decided on a five-day merge freeze before each release. Only bug fixes could go in.<\/p>\n<p>It wasn&#8217;t popular. Jonas, our most senior engineer, thought it would just move the rush earlier. Our product manager, Ines, had promised a feature to marketing for the next release. They both had a point. So I changed one thing. Features could still merge during the freeze behind a flag that stayed off. That way code landed early and only the launch waited.<\/p>\n<p>Before the first freeze, I put in our planning doc when we&#8217;d drop it. It would go if crash-free sessions didn&#8217;t improve over three releases. It would also go if more than one planned feature slipped per quarter. I asked Jonas to run the review, since he was the one most against it.<\/p>\n<p>After three releases, crash-free sessions went from 98.9% to 99.6%. We had no hotfixes. One feature slipped. Jonas also found that people were merging big changes the day before the freeze. So he was partly right. We cut the freeze to three days and added a size limit on merges just before it. I&#8217;d make the same call again. I&#8217;d also set the review date the same way.<\/p>\n<\/blockquote>\n<p>The answer never claims the team came around or that Tomas was proven right. It says what he decided, what the critics were right about and what the review found. Our post on <a href=\"https:\/\/grindloop.ai\/blog\/tell-me-about-a-time-you-said-no\/\">saying no to someone who outranked you<\/a> covers the case where you lose the argument. Did your call overrule your manager? Then see <a href=\"https:\/\/grindloop.ai\/blog\/disagreed-with-your-manager-interview\/\">why a &#8220;disagreed with your manager&#8221; answer stops at the decision<\/a>.<\/p>\n<h2>If you never set a reversal condition, name the evidence that would have changed your mind<\/h2>\n<p>Many real decisions had no written exit. You made the call and the team moved on. Don&#8217;t add a condition after the fact. A follow-up like &#8220;where did you write that down?&#8221; will expose it.<\/p>\n<p>Say instead what evidence would have made you reverse it. Then say whether you looked for it. Tomas&#8217;s version without a planning doc might go like this. He&#8217;d say he watched crash-free sessions and planned slips after each release. He&#8217;d say he would have dropped the freeze if either got worse. Then he&#8217;d add what he does now. He writes the reversal condition into the decision before it starts. He also gives the review to the person who argued hardest against it. That last part also answers &#8220;what would you do differently?&#8221; before the interviewer asks it.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>An unpopular decision interview answer names what the critics got right and the condition that would reverse your call. Full Android sample answer inside.<\/p>\n","protected":false},"author":4,"featured_media":949,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"rank_math_title":"Why Your Unpopular Decision Interview Answer Sounds Like You Won the Argument","rank_math_description":"An unpopular decision interview answer names what the critics got right and the condition that would reverse your call. Full Android sample answer inside.","rank_math_focus_keyword":"unpopular decision interview answer","footnotes":""},"categories":[10],"tags":[35,16,22,19],"class_list":["post-947","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-behavioral","tag-behavioral-interview","tag-interview-prep","tag-senior-interview","tag-soft-skills"],"_links":{"self":[{"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/posts\/947","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/comments?post=947"}],"version-history":[{"count":1,"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/posts\/947\/revisions"}],"predecessor-version":[{"id":948,"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/posts\/947\/revisions\/948"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/media\/949"}],"wp:attachment":[{"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/media?parent=947"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/categories?post=947"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/grindloop.ai\/blog\/wp-json\/wp\/v2\/tags?post=947"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}