{"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":"2026-08-24T22:57:06.136Z"},"content":[{"type":"blog","id":"747497fa-ee9c-411d-8bcc-8fdcfd696cd8","slug":"government-customer-research-paperwork-reduction-act-2026","title":"Government Customer Research in 2026: The Ten-Person Rule That Triggers Federal Approval","url":"https://www.koji.so/blog/government-customer-research-paperwork-reduction-act-2026","summary":"Under 5 CFR 1320.3(c) a federal agency, or a vendor collecting for one, needs OMB approval before posing identical questions to ten or more persons in any 12-month period. Voluntary customer satisfaction surveys are covered. Clearance requires a 60-day Federal Register notice, a 30-day notice on submission, and up to 60 days of OMB review, and lasts no longer than three years.","content":"## The short answer\n\nIf you are a federal agency, or a vendor collecting on one's behalf, asking the same question of ten or more people within a twelve-month period is a \"collection of information\" under the Paperwork Reduction Act. It needs approval from the Office of Management and Budget before you field it. Making the survey voluntary does not exempt it. Calling it customer experience research does not exempt it. The Congressional Research Service names the exact case in its overview of the Act: collections subject to the PRA include \"collections that are required to obtain or retain a benefit, and voluntary collections, such as customer service satisfaction surveys.\"\n\nMost product and research teams discover this after they have written the questionnaire. This guide explains where the line actually falls, what clearance costs in calendar time, and how to design a public sector research program that survives the process instead of being redesigned by it.\n\n## What the law actually regulates\n\nThe operative definition lives in the PRA's implementing regulation at 5 CFR 1320.3(c). A collection of information means:\n\n> the obtaining, causing to be obtained, soliciting, or requiring the disclosure to an agency, third parties or the public of information by or for an agency by means of identical questions posed to, or identical reporting, recordkeeping, or disclosure requirements imposed on, ten or more persons, whether such collection of information is mandatory, voluntary, or required to obtain or retain a benefit.\n\nFour things in that sentence decide whether your study is regulated, and three of them surprise people.\n\n**\"By or for an agency.\"** The obligation follows the sponsor, not the fieldworker. If an agency funds, directs or sponsors the work, a contractor running the interviews is inside the regime. Hiring an agency-of-record or a research platform does not move the study outside it.\n\n**\"Identical questions.\"** The trigger is standardization, not curiosity. A discussion guide that asks every participant the same thing is the regulated artifact. This is the single most consequential word in the definition, and it has consequences we cover in detail in our companion piece on [the PRA's exemption structure](/blog/paperwork-reduction-act-research-exemptions-2026).\n\n**\"Ten or more persons.\"** Not ten responses, and not ten per study. Section 1320.3(c)(4) defines the phrase as \"the persons to whom a collection of information is addressed by the agency within any 12-month period.\" Inviting a hundred people and getting nine replies is still a collection addressed to a hundred people. Splitting one study into batches of nine across a quarter does not work either, because the window is twelve months, not per-wave.\n\n**\"Mandatory, voluntary, or required to obtain or retain a benefit.\"** All three are covered. The voluntariness of your survey is a fact about respondent burden, not about your obligations.\n\nThe Act is codified at 44 U.S.C. sections 3501 to 3521, and it applies, per CRS, to \"almost all executive branch agencies, including statutorily designated independent regulatory agencies.\"\n\n## The threshold is legal, not statistical\n\nThis is the part worth sitting with, because it inverts an instinct that every researcher has been trained on.\n\nIn every other context, sample size is a statistical question. You choose n to hit a confidence interval, to reach thematic saturation, or to fill a segment grid. Under the PRA, ten is a jurisdictional boundary. Nine is unregulated. Ten is regulated. The number carries no statistical meaning at all, and it does not scale with the sensitivity of what you are asking or the burden you are imposing. A three-question satisfaction poll sent to ten people is covered. A two-hour unstructured conversation with nine people is not.\n\nThe practical consequence is that public sector research programs get designed around a number that has nothing to do with evidence quality. Teams that do not know the rule blow through it without noticing. Teams that do know it often shrink studies to stay under it, which is a real methodological cost paid for a purely administrative reason.\n\n## What clearance actually costs in calendar time\n\nThe regulation lays out a sequence, and the sequence is the schedule. Under 5 CFR 1320.8(d)(1), before an agency submits a collection to OMB it \"shall provide 60-day notice in the Federal Register, and otherwise consult with members of the public and affected agencies concerning each proposed collection of information.\" Then, under 5 CFR 1320.10(a), on or before the date it submits to OMB the agency publishes a second notice requesting comments to OMB \"within 30 days of the notice's publication.\" OMB then has, under 1320.10(b), sixty days from receipt or publication, whichever is later, to approve, require changes, or disapprove.\n\n| Stage | Governing section | Statutory clock |\n| --- | --- | --- |\n| First Federal Register notice | 5 CFR 1320.8(d)(1) | 60 days of public comment |\n| Second notice on submission to OMB | 5 CFR 1320.10(a) | 30 days of comment to OMB |\n| OMB decision | 5 CFR 1320.10(b) | up to 60 days from the later of receipt or publication |\n| Approval lifetime | 5 CFR 1320.10(b) | maximum of three years |\n| Emergency processing | 5 CFR 1320.13(f) | control number valid a maximum of 90 days |\n\nAdd the agency's own internal review before anything reaches the Federal Register and a first-time clearance is a multi-quarter project. Note the last two rows in particular. An approval expires: OMB \"shall not approve any collection of information for a period longer than three years,\" so a standing research program is a renewal treadmill, not a one-time cost. And the emergency valve under 1320.13 is genuinely narrow. It requires a written determination that public harm is likely, that an unanticipated event occurred, or that normal procedures would cause a deadline to be missed, and what it buys is a control number good for ninety days.\n\nThere is one protection running the other way, and researchers should know it because respondents sometimes ask. Section 1320.6 provides that \"no person shall be subject to any penalty for failing to comply with a collection of information\" that does not display a currently valid OMB control number. That is why cleared federal instruments carry a control number and an expiration date on the face of the form.\n\n## The burden estimate is the research design\n\nHere is the mechanism almost nobody anticipates, illustrated with a live example.\n\nOn 20 August 2026 the National Science Foundation published a 60-day notice in the Federal Register, document 2026-16984, renewing its customer experience collection under OMB Circular A-11 Section 280. Comments are due 19 October 2026. The notice carries OMB Clearance Number 3145-0254 and two numbers that define the program: an Estimated Number of Respondents of 2,001,550, and Estimated Total Annual Burden Hours of 101,125.\n\nDivide one by the other. The notice commits to an average of just over three minutes per respondent. The document itself says response time is \"varied,\" and that it \"may be 3 minutes or up to 2 hours to participate in an interview.\" But the arithmetic of the total tells you what mix was actually assumed. If every respondent other than the interviewees takes three minutes, the remaining budget supports roughly 537 two-hour interviews across the entire program, which is about 0.03% of respondents.\n\nThat is not a criticism of NSF, whose notice is a routine and properly filed renewal. It is a structural observation about how the regime shapes method. **The burden estimate is written to justify the clearance, and it then functions as a cap on how much qualitative work the program can do.** A number chosen as a compliance artifact silently pre-commits the agency to a survey-shaped research portfolio, because depth is expensive in burden hours and breadth is cheap.\n\nTwo more details from that notice are worth carrying into your own planning. First, NSF states it \"will limit its inquiries to data collections that solicit strictly voluntary opinions or responses,\" and it is still filing for clearance, which is the voluntariness point made concrete. Second, the covered methods are explicitly qualitative as well as quantitative, including \"interviews, questionnaires, surveys, and focus groups,\" with outputs described as \"the creation of personas, customer journey maps and reports.\" Ordinary user research artifacts, cleared through a federal paperwork process.\n\n## Scale, and why the rule exists\n\nThe burden the PRA governs is not notional. Reporting the figures OMB publishes in its annual information collection budget, CRS records that \"in FY2022, the total paperwork burden reported by OMB was 10.34 billion hours, compared to 9.97 billion hours in FY2021.\" Ten billion hours is on the order of five million full-time work-years, which is why a statute exists to make agencies justify each question they ask.\n\nUnderstanding that is what makes the 60-day notice tractable rather than adversarial. The comment period exists to test four things named in 1320.8(d)(1): whether the collection is necessary and has \"practical utility,\" whether the burden estimate is accurate, whether the instrument is clear, and whether burden can be minimized \"including through the use of appropriate automated, electronic, mechanical, or other technological collection techniques.\" That last clause is an explicit invitation to argue that a better instrument reduces burden.\n\n## What is genuinely outside the regime\n\nSection 1320.3(h) lists categories that are generally not \"information,\" and 1320.4 excludes collections during federal criminal investigations and certain intelligence activities. The ones that matter for research teams are:\n\n- **A request addressed to a single person** (1320.3(h)(6)). One interview, one person, no clearance.\n- **Facts or opinions obtained through direct observation** by an agency employee or agent, or through \"nonstandardized oral communication in connection with such direct observations\" (1320.3(h)(3)).\n- **Nonstandardized follow-up questions** designed to clarify responses to an already-approved collection (1320.3(h)(9)).\n- **Facts or opinions obtained or solicited at or in connection with public hearings or meetings** (1320.3(h)(8)).\n- **General solicitations of comments from the public** published in the Federal Register (1320.3(h)(4)), provided respondents are not required to supply information beyond self-identification.\n\nRead that list carefully and a pattern emerges: every exemption is an exemption from standardization. That is the subject of the companion article, and it is where most of the real methodological risk lives.\n\n## Running public sector research that survives the process\n\nCompliance and speed are not actually opposed here, but you have to design for the regime rather than around it.\n\n**Clear one broad instrument, not many narrow ones.** Circular A-11 Section 280 collections like NSF's exist precisely so an agency can run many customer feedback activities under a single umbrella clearance. Scope the umbrella wide and generic at the start.\n\n**Write the burden estimate as a research budget.** If you want depth later, buy the hours now. Modeling a realistic share of long-form interviews at submission is far cheaper than a change request eighteen months in.\n\n**Use the approved core plus lawful adaptive probing.** A cleared, standardized question set that every respondent receives, with nonstandardized clarifying follow-ups layered on top, is the architecture the regulation actually contemplates at 1320.3(h)(9).\n\nThat last pattern is exactly what an AI-moderated interview does well, and it is where Koji fits a public sector program. Koji studies are built from six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, and yes_no), so the standardized core of your instrument is explicit, reviewable, and reproducible for every participant, which is what a supporting statement needs to describe. The AI interviewer then probes conversationally on top of that fixed core, so you get interview-grade depth without a different question set for each respondent. See our guides to [structured questions in AI interviews](/docs/structured-questions-guide) and [how AI-moderated interviews work](/docs/ai-moderated-interviews) for the mechanics.\n\nBecause interviews run asynchronously and analysis is automatic, the depth that used to be unaffordable in burden hours becomes practical: a five-minute voice interview yields far more than a five-minute form, at the same burden cost per respondent. Thematic analysis and reports are generated without a moderator scheduling a single call, which is what turns a cleared instrument into a program you can actually run at agency scale rather than a form you field once a year.\n\nFor the surrounding compliance work, our documentation covers [recording consent laws](/docs/interview-recording-consent-laws), [intake forms and consent](/docs/intake-forms-and-consent), [enterprise security review](/docs/enterprise-security-ai-research-platforms), and the [legal rules on research recruitment outreach](/docs/research-recruitment-outreach-law).\n\n## Frequently asked questions\n\n### Does the Paperwork Reduction Act apply to voluntary surveys?\n\nYes. 5 CFR 1320.3(c) covers collections \"whether such collection of information is mandatory, voluntary, or required to obtain or retain a benefit.\" The Congressional Research Service specifically lists \"voluntary collections, such as customer service satisfaction surveys\" as subject to the Act. Voluntariness affects how you describe burden to respondents, not whether clearance is required.\n\n### Can I avoid clearance by surveying only nine people?\n\nOnly if you genuinely address the instrument to fewer than ten people in a twelve-month period. Section 1320.3(c)(4) defines \"ten or more persons\" as those \"to whom a collection of information is addressed by the agency within any 12-month period,\" so running repeated waves of nine within a year does not work. It also counts people you invited, not people who answered.\n\n### Does the PRA apply to contractors and research vendors?\n\nThe definition covers information obtained \"by or for an agency,\" so a contractor collecting on an agency's behalf is inside the regime. The clearance obligation sits with the sponsoring agency, but a vendor that fields an uncleared instrument creates a compliance problem for its client. Confirm the OMB control number before fielding.\n\n### How long does OMB clearance take?\n\nThe statutory minimum sequence is a 60-day Federal Register notice under 5 CFR 1320.8(d)(1), then a 30-day comment window on submission under 1320.10(a), with OMB having up to 60 days to decide under 1320.10(b). Agency-internal review comes on top. Approvals last no longer than three years, so renewals recur.\n\n### What is an OMB control number and why is it on the form?\n\nIt is the identifier OMB assigns on approval. Under the public protection provision at 5 CFR 1320.6, no person may be penalized for failing to comply with a collection that does not display \"a currently valid OMB control number,\" which is why cleared federal instruments print the number and an expiration date.\n\n### Is emergency processing a realistic shortcut?\n\nRarely. Section 1320.13 requires a written determination that the collection is essential to the agency mission and that normal procedures cannot be followed because public harm is likely, an unanticipated event occurred, or a statutory or court-ordered deadline would be missed. If granted, the control number is valid for a maximum of 90 days.\n\n## Run cleared research at agency speed with Koji\n\nThe Paperwork Reduction Act does not stop you doing excellent public sector research. It just means the standardized core of your instrument has to be written down, justified, and lived with for up to three years, so the quality of that core matters more than in any commercial program.\n\nKoji is built for exactly that shape of work: an explicit, reviewable set of structured questions that every participant receives identically, plus an AI interviewer that probes conversationally on top of it. You get interview depth at survey scale, automatic thematic analysis, and one-click reports, with no moderator bias and no scheduling. From question to insight in hours, not weeks, and no research expertise required to run it.\n\n[Start a free Koji study](https://www.koji.so) and see what a cleared instrument can do when the depth is not capped by moderator time.","category":"Research","lastModified":"2026-08-24T03:33:11.603882+00:00","metaTitle":"Government Customer Research and the Paperwork Reduction Act (2026)","metaDescription":"Ten or more people, one identical question, one federal agency: that is a PRA collection needing OMB approval. The threshold, the timeline, and how to design for it.","keywords":["government customer research","public sector user research","paperwork reduction act survey","omb clearance research","federal customer experience research","pra ten or more persons","government survey approval"],"aiSummary":"Under 5 CFR 1320.3(c) a federal agency, or a vendor collecting for one, needs OMB approval before posing identical questions to ten or more persons in any 12-month period. Voluntary customer satisfaction surveys are covered. Clearance requires a 60-day Federal Register notice, a 30-day notice on submission, and up to 60 days of OMB review, and lasts no longer than three years.","aiKeywords":["paperwork reduction act","omb control number","information collection request","public sector research","federal register notice","burden hours"],"aiContentType":"guide","faqItems":[{"answer":"Yes. 5 CFR 1320.3(c) covers collections \"whether such collection of information is mandatory, voluntary, or required to obtain or retain a benefit.\" The Congressional Research Service specifically lists \"voluntary collections, such as customer service satisfaction surveys\" as subject to the Act. Voluntariness affects how you describe burden to respondents, not whether clearance is required.","question":"Does the Paperwork Reduction Act apply to voluntary surveys?"},{"answer":"Only if you genuinely address the instrument to fewer than ten people in a twelve-month period. Section 1320.3(c)(4) defines \"ten or more persons\" as those \"to whom a collection of information is addressed by the agency within any 12-month period,\" so running repeated waves of nine within a year does not work. It also counts people you invited, not people who answered.","question":"Can I avoid clearance by surveying only nine people?"},{"answer":"The definition covers information obtained \"by or for an agency,\" so a contractor collecting on an agency's behalf is inside the regime. The clearance obligation sits with the sponsoring agency, but a vendor that fields an uncleared instrument creates a compliance problem for its client. Confirm the OMB control number before fielding.","question":"Does the PRA apply to contractors and research vendors?"},{"answer":"The statutory minimum sequence is a 60-day Federal Register notice under 5 CFR 1320.8(d)(1), then a 30-day comment window on submission under 1320.10(a), with OMB having up to 60 days to decide under 1320.10(b). Agency-internal review comes on top. Approvals last no longer than three years, so renewals recur.","question":"How long does OMB clearance take?"},{"answer":"It is the identifier OMB assigns on approval. Under the public protection provision at 5 CFR 1320.6, no person may be penalized for failing to comply with a collection that does not display \"a currently valid OMB control number,\" which is why cleared federal instruments print the number and an expiration date.","question":"What is an OMB control number and why is it on the form?"},{"answer":"Rarely. Section 1320.13 requires a written determination that the collection is essential to the agency mission and that normal procedures cannot be followed because public harm is likely, an unanticipated event occurred, or a statutory or court-ordered deadline would be missed. If granted, the control number is valid for a maximum of 90 days.","question":"Is emergency processing a realistic shortcut?"}],"relatedTopics":["enterprise customer research","HCP panels","Research Incentives","public sector research"]}],"pagination":{"total":1,"returned":1,"offset":0}}