{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-09-24T13:21:35.624Z"},"content":[{"type":"documentation","id":"9e5dc536-fc45-41d4-9be8-3b5e29363944","slug":"sterile-cockpit-rule-interviews","title":"The Sterile Cockpit Rule for User Interviews: When to Stop Making Conversation","url":"https://www.koji.so/docs/sterile-cockpit-rule-interviews","summary":"Imports aviation's sterile cockpit rule (14 CFR 121.542, prompted by the 1974 Eastern Air Lines Flight 212 crash) into interview moderation. NASA ASRS reviewed 63 relevant reports: 48% altitude deviations, 14% course deviations, 14% runway transgressions, 14% general distraction, 8% takeoff or landing without clearance, 2% near mid-air collision. Key correction: \"sterile\" is phase-scoped, not cold - the rule schedules rapport rather than banning it. Maps flight phases onto interview phases and names four critical phases (episodic recall, the first substantive answer, genuine hesitation, the follow-up after a surprising statement). The research analogue of an altitude deviation is losing your place in the guide. Also covers the observation-room problem as the equivalent of a cabin call during approach, and why offering an example during a pause is the most destructive violation.","content":"Aviation solved a version of your interview problem in 1981, and the solution has a name. Below 10,000 feet, crews are forbidden from talking about anything that is not flying the aircraft. The rule exists because a specific accident proved that friendly, harmless, well-intentioned conversation kills people during the phases of flight that demand full attention.\n\nA research interview has critical phases too. Almost nobody names them, so almost nobody protects them - and the default moderator habit is to be warmest and chattiest at exactly the moment the participant is doing the hardest cognitive work.\n\n## Where the rule comes from\n\nIn 1974, Eastern Air Lines Flight 212 crashed short of the runway at Charlotte during an instrument approach in fog. The NTSB found that a probable cause was lack of altitude awareness caused by distraction from idle chatter among the crew during the approach. Nobody was negligent in any ordinary sense. They were chatting.\n\nThe FAA's response became 14 CFR 121.542. NASA's Aviation Safety Reporting System reviewed the reports that fed the rulemaking, and its own summary states it plainly: \"Commonly known as the 'sterile cockpit rule,' these regulations specifically prohibit crew member performance of non-essential duties or activities while the aircraft is involved in taxi, takeoff, landing, and all other flight operations conducted below 10,000 feet MSL, except cruise flight.\"\n\n### What the regulation actually prohibits\n\nThe regulation is specific about the behaviours it bans, naming among them \"eating meals, engaging in nonessential conversations within the cockpit\" and reading material unrelated to the flight. Note what is on that list: not misconduct, not incompetence, not fatigue. Ordinary, sociable, human behaviour that is entirely appropriate half an hour earlier.\n\nThat is the part worth importing. The rule does not regulate bad behaviour. It regulates good behaviour occurring at the wrong time.\n\n### What the NASA data says the rule prevents\n\nThe ASRS analysis reviewed 63 incident reports coded as relevant to the sterile cockpit rule. The consequence breakdown is the useful part:\n\n| Consequence | Share of reports |\n| --- | --- |\n| Altitude deviations | 48% |\n| Course deviations | 14% |\n| Runway transgressions | 14% |\n| General distraction, no specific consequence | 14% |\n| Takeoff or landing without clearance | 8% |\n| Near mid-air collision | 2% |\n\nOne pilot's report, quoted in the NASA analysis, is the whole argument in a sentence: \"If we [had] adhered to the sterile cockpit, this situation probably would not have occurred.\"\n\nNearly half of all consequences were altitude deviations. The dominant failure mode of distraction is not a dramatic error. It is losing track of where you are.\n\n### The mistranslation: sterile does not mean cold\n\nThe word does a lot of damage. Importing this rule badly produces a moderator who is clipped, transactional, and unpleasant for a full hour, which reliably produces worse data than chattiness ever did.\n\nThe rule is not \"be cold.\" It is \"non-essential activity is banned during defined phases, and welcome outside them.\" Crews are not silent in cruise. They talk about the weather, their kids, and the catering. The rule is phase-scoped, and the phases are short.\n\nRapport-building is not the opposite of this rule. It is the activity the rule schedules.\n\n### Mapping flight phases onto an interview\n\n| Flight phase | Interview phase | Sterile? |\n| --- | --- | --- |\n| Taxi and pre-departure | Greeting, consent, setup, expectation-setting | No - build rapport here |\n| Takeoff and climb | The first substantive question | Yes |\n| Cruise | Mid-interview exploration, broad topics | No - warmth is fuel here |\n| Descent and approach | Critical incident recall, the question the study exists for | Yes |\n| Landing and taxi in | Closing, thanks, next steps | No |\n\nThe insight this produces is not obvious and is worth stating directly: the sterile phases are the beginning and the end of the data, not the beginning and the end of the session.\n\n### The critical phases of a research interview\n\nA phase is critical when the participant is doing work that interruption destroys rather than merely delays. In practice that means:\n\n- Episodic recall. The participant is reconstructing a specific past event, which is fragile and effortful, and a friendly interjection collapses it.\n- The first substantive answer. Whatever frame the participant adopts here tends to govern the rest of the session.\n- Any moment of genuine hesitation. A pause before an answer is frequently the participant assembling something they have never said out loud.\n- The follow-up to a surprising statement. This is the highest-value seam in the whole interview and the easiest to fill with noise.\n\n### Why moderators violate the rule exactly when it matters\n\nSilence is uncomfortable, and the discomfort peaks precisely when the participant goes quiet to think. The moderator reads the pause as a failure of the question and rushes in to rescue it - rephrasing, offering an example, filling the gap with an anecdote of their own.\n\nEvery one of those rescues is a nonessential activity below 10,000 feet. The participant was not stuck. They were working.\n\nOffering an example is the most damaging version, because it does not merely interrupt the recall; it supplies the content the participant was about to generate, and you can no longer tell whose answer you recorded.\n\n### The altitude-deviation analogue\n\nForty-eight percent of sterile cockpit incidents were altitude deviations - crews losing track of where they were. The interview equivalent is losing your place in the guide, and it shows up in three recognisable forms:\n\n- Skipping a question and only noticing in analysis, by which point the participant is gone.\n- Asking the same question twice, which signals inattention and costs you rapport you were trying to protect.\n- Landing early. Running out of guide with fifteen minutes left because you rushed the middle while chatting, then padding with low-value questions.\n\nThe cheap fix is the same one aviation uses: a written altitude call. Know which question you are on, out loud in your own head, at every transition.\n\n### The observer problem: your cockpit has passengers\n\nThe regulation also restricts the crew from being distracted by non-flying duties, including communications from outside the flight deck. Research has an exact analogue that most teams have simply accepted: stakeholders in the observation room messaging the moderator mid-session.\n\nA stakeholder typing \"can you ask about pricing?\" during a participant's critical-incident recall is the research equivalent of a cabin call during approach. It is well-intentioned, it is relevant, and it arrives at the worst possible moment.\n\nThe fix is not to exclude observers. It is to schedule their input: questions are collected during cruise phases and at the end, never during a sterile phase. Say this out loud before the session, the way a crew briefs an approach.\n\n## How Koji handles this\n\n- Koji's AI interviewer does not get uncomfortable with silence. It does not rescue a thinking participant, does not offer an example to fill a pause, and does not supply the answer the participant was assembling - which is the single most destructive sterile-phase violation.\n- Koji's AI follow-up questions arrive after the participant has finished, not over the top of them, so the highest-value seam in the interview stays clean.\n- Koji's structured questions enforce the altitude call by construction. With six types - open_ended, scale, single_choice, multiple_choice, ranking, and yes_no - carrying stable identifiers through the session, no question is silently skipped and none is asked twice.\n- Koji removes the observation-room channel entirely. Stakeholders read real-time reports instead of messaging a moderator mid-session, so their questions inform the next study rather than derailing this one.\n- Koji runs voice and text interviews without a moderator present, which means the sterile phases are sterile by default rather than by discipline.\n\nWorth being straight about the limit: this rule is mostly about moderator behaviour, so an unmoderated platform sidesteps it rather than teaching it. If you run moderated sessions, you still need the discipline - and the phase map above is the thing to put on the wall.\n\n## Common mistakes\n\n- Reading \"sterile\" as \"cold\" and running the entire session without warmth, which costs more data than chattiness does.\n- Treating the whole interview as critical, which is unsustainable and makes the rule collapse within two sessions.\n- Filling a thinking pause with an example, which substitutes your content for the participant's.\n- Letting observers interrupt on their schedule rather than yours.\n- Putting the highest-value question last, when both moderator attention and participant energy are lowest.\n\n## Frequently asked questions\n\n### Does the sterile cockpit rule mean I should stop building rapport?\n\nNo, and this is the most common misreading. The rule is phase-scoped, not global. Crews talk freely in cruise and stop below 10,000 feet. Applied to an interview, rapport-building belongs in the opening, in transitions, and in the close, and it is actively harmful only during critical phases such as episodic recall or the first substantive answer. The rule schedules warmth rather than banning it.\n\n### Which parts of a user interview count as critical phases?\n\nA phase is critical when interruption destroys the participant's work rather than delaying it. In practice: episodic recall of a specific past event, the first substantive answer of the session because it frames everything after it, any hesitation that indicates the participant is assembling a thought they have not said before, and the follow-up immediately after a surprising statement. Those four cover nearly all of the value in a typical session.\n\n### What is the research equivalent of an altitude deviation?\n\nLosing your place in the guide. Altitude deviations were 48% of the sterile cockpit incidents NASA reviewed, and the interview version is just as common: skipping a question and discovering it in analysis, asking the same question twice, or running out of guide early because the middle was rushed. All three come from the same cause, which is attention spent on something other than where you currently are.\n\n### How do I stop stakeholders interrupting during sessions?\n\nBrief them the way a crew briefs an approach. Tell observers in advance that questions are collected during broad exploratory stretches and at the end, and that the critical-recall portion is closed. The goal is to schedule their input rather than exclude it. Koji avoids the problem structurally by giving stakeholders real-time reports to read instead of a moderator to message.\n\n### Is silence really better than helping a participant who seems stuck?\n\nUsually, yes - because most participants who look stuck are working rather than blocked. The tell is what you do to help. Repeating the question verbatim is safe. Offering an example is not, because it hands the participant the content they were about to generate and you can no longer attribute the answer. Wait roughly twice as long as feels comfortable before intervening at all.\n\n### Does an AI interviewer make this rule unnecessary?\n\nIt sidesteps the moderator half of it. Koji does not find silence uncomfortable, so it does not rescue a thinking participant or fill a pause with a leading example, and it does not lose its place in the guide. What it does not do is teach the discipline to a human moderator. If your team runs moderated sessions alongside automated ones, the phase map still has to be learned.\n\n## Related Resources\n\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide) - stable question identifiers make a skipped or repeated question impossible\n- [Building Rapport in Research Interviews](/docs/building-rapport-interviews) - the activity this rule schedules rather than forbids\n- [User Interview Guide Template](/docs/user-interview-guide-template) - where to mark the critical phases in your own guide\n- [Recall Bias](/docs/recall-bias) - why episodic recall is the most fragile phase to interrupt\n- [Avoiding Bias in Research Interviews](/docs/avoiding-bias-in-interviews) - what happens when you supply the answer during a pause\n- [Cognitive Biases in User Interviews](/docs/cognitive-biases-user-interviews) - the broader set of moderator-introduced distortions","category":"Interview Techniques","lastModified":"2026-09-24T03:28:12.236031+00:00","metaTitle":"The Sterile Cockpit Rule for User Interviews (2026)","metaDescription":"Aviation bans non-essential talk below 10,000 feet. Interviews have critical phases too - and moderators are chattiest at the worst moment.","keywords":["sterile cockpit rule interviews","interview distractions","when to stop small talk interview","critical phase interview","moderator silence","interview observation room"],"aiSummary":"Imports aviation's sterile cockpit rule (14 CFR 121.542, prompted by the 1974 Eastern Air Lines Flight 212 crash) into interview moderation. NASA ASRS reviewed 63 relevant reports: 48% altitude deviations, 14% course deviations, 14% runway transgressions, 14% general distraction, 8% takeoff or landing without clearance, 2% near mid-air collision. Key correction: \"sterile\" is phase-scoped, not cold - the rule schedules rapport rather than banning it. Maps flight phases onto interview phases and names four critical phases (episodic recall, the first substantive answer, genuine hesitation, the follow-up after a surprising statement). The research analogue of an altitude deviation is losing your place in the guide. Also covers the observation-room problem as the equivalent of a cabin call during approach, and why offering an example during a pause is the most destructive violation.","aiDifficulty":"intermediate","aiEstimatedTime":"9 min"}],"pagination":{"total":1,"returned":1,"offset":0}}